With the recent two month long stretch of clouds and rain, I decided to tackle the number one issue causing me to not take advantage of the rare (lately) clear sky…setup time. Living in the suburbs of Houston with a tiny yard, I don’t have the option of building a permanent, back-yard observatory. My best location is right in the middle of my driveway and even then I only have about 70 degrees of visible sky east to west…but I do have a clear shot of Polaris. The problem was that setting up is an hour long, very physical process and I can’t leave my gear in the driveway all night so I have to tear down at 2-3 in the morning…so…I would talk myself out of it most nights. If I wanted to image more frequently, I needed to simplify my setup/teardown routine!
Step one was easy and the most important. Getting a heavy duty medium size wheeley bar from JMI allowed me to leave the scope setup and just roll it in and out each night. This alone saved 45 minutes of labor!
After using this setup for a while, I realized that I was still spending 15 minutes running cables to the computer and power to all the electronics so I started looking at on-telescope computer control systems. I did find two commercial options. The AstroBar was light weight and reasonably priced but feature and performance poor. The Eagle2 had the performance and many of the features I wanted but it was fairly heavy and very expensive. Having electrical experience, a couple of 3D printers, and plenty of cloudy nights, I began experimenting with stick PCs. The results were generally positive but I did run into a few issues. First, I found most of the sticks I tested to be fairly underwhelming from a performance standpoint. Running Windows 10, camera control, image capture, guiding software, etc. on 2 GB of RAM with an Atom processor proved troublesome at times. Plus, there were multiple occasions where I lost my remote desktop session to the stick and I had to drag a monitor, keyboard, and mouse out to the scope to plug in the stick and see what what going on. The first time this happened, I was at the Houston Astronomical Society dark site with no peripherals…3 hour round trip plus setup/tear-down time for no images. Very frustrating! Additionally, at this point, the cable monster was beginning to get a bit out of hand! The telescope looked like it had lost a fight with a drunk Radio Shack sales associate!
It was time to plan my ideal, custom Astrophotography, on-telescope computer/control setup.
Requirements:
- Incorporate the following products into one system
- Computer
- Dew Control
- Camera Control
- USB Hub and Power
- 12v Accessory Power
- Provide at least 6 USB 3.0 ports (primary camera data, guide camera, PoleMaster, mount control, hand controller if doing PEC, focuser, one extra)
- Power status monitor
- Small integrated touch screen display (back up to lost remote sessions)
- Single cable running off scope (power)
- Small
- LIGHT WEIGHT
After two months of prototyping and testing I am very pleased with the finial product…I call it The Cube 
Specs:
- Weight – 982g/2.16 lbs (this is actually 74g less on-scope than my previous non-integrated setup)
- Size – 16cm/6.25 inch cube
- Intel M5 CPU, 4GB of RAM, 256GB of storage
- 4 channel variable dew control w/low voltage cutoff
- 7 external USB 3.0 ports each delivering up to 2.4 amps
- 3 internal powered USB 3.0 ports
- 5 inch multi-touch LCD display
- Real-time voltage/amperage monitor
- Highly regulated power control for the compute stick
- QHY 10 camera control (TEC cooling/power/status)
- Switched control for USB Hub, PC, Dew Control, and QHY CCD Camera
- 4 unswitched 12v 5.5×2.1 mm accessory power jacks
Design and construction
I began the process by taking all the separate components I was trying to integrate and attaching them to a wooden block that could mount on the telescope so I could prove the concept and get a feel for the wiring involved and any potential pitfalls.
Next, I 3D printed some components that would let me mock up the internal spacing requirements. The USB hub I decided to use based on size and weight was wedge shaped. I had to measure the angle and print a test wedge (orange component below) to stand the hub vertical for the mockup. The yellow stand below straddles the power distribution block and the buck converter used to supply the correct voltage (5v) to the USB hub. The red box houses the compute stick power regulator (more on that below) and the green shelf mocks up the location of the compute stick complete with opening for the cooling fan.
The next step was to layout the front and rear panels. The front panel contains the USB hub, individual power switches (hub, PC, dew control, and QHY10 camera), PC power regulator control, master power meter, and dew heater outputs. The rear panel contains the touch screen and dew heater controls.
As a starting point I found a great parametric box and panel maker script that allowed me to use OpenSCAD to define my panel layouts mathematically after careful measurements with an accurate set of digital calipers.
This method let me rapidly print and test prototypes for fit and layout.
Time for the real design work. Using the prototype prints from OpenSCAD as a starting point, I used SketchUp Pro 2018 to design the final box and panels. The front and rear panels feature raised lettering which allowed me to change filament colors once that layer was reached during printing (I went with light grey panels, orange lettering, and a black box to match my Celestron RASA/CGX-L setup). The upper box has an inset detail with precise holes for the QHY DC-201 status LEDs (more on that below). The lower box has a built-in wedge to lock in the USB hub in the correct position as well as built in tabs for positioning and holding the power distribution block and buck converter. Additionally, the compute stick shelf and integrated latch has a dovetail which joins it to the lower box half.
Printing time!
Final parts ready for assembly.
One of the trickiest aspects of the build was powering the Intel STK2MV64CC Compute Stick. When I tested earlier Atom based compute sticks, all I had to do was wire up a mini USB plug to a 5V power supply and the stick fired right up. However, I wanted a stick with a full m5 CPU and 4Gb of RAM which meant going with the USB-C powered version. The first thing I did measure the input voltage and current with a USB-C multimeter.
Thinking I was in good shape with 5V, I tried a simple 12V to 5V USB-C converter.
The stick would power-up, enter the BIOS check screen, then shut down…odd.
So I tried a higher end buck converter…same result…power-up, BIOS check, shut down.
So, I decided to take apart the AC power adapter that came with the stick to see how it worked. This unique adaptor plugs into the wall and then delivers 5V over USB-C. Additionally, the wall adapter itself has two additional USB 3.0 ports that act as a hub over the type C connection. (the stick only has one 3.0 port on-board).
Inside, I found a normal 110V AC to 5V DC convertor attached to a daughter board containing the USB 3.0 ports and USB-C connection.
Measuring the output signal on the USB-C link, I found a non-standard signal embedded in the normal data channel. It seems that Intel decided to stop users from using any power supply except the supplied Intel version. If the BIOS does not detect this input signal on the data channel of the USB-C connection, it shuts down. There is no way around it. I guess they were tired of users burning up their compute sticks with crappy cell phone chargers or other poorly controlled power supplies. My options at this point were to run a LONG USB-C cable along with the 12V power cable down from the scope to the ground and use the Intel power supply…or…get creative!
I realized that the daughter board was the only component that mattered and it was very small and light weight. The bulk of the power supply was the 110V to 5v converter which I didn’t need. I already had an extremely accurate voltage regulator in the Cube which I was planning on using to supply 5V directly to the USB-C port on the compute stick. Now I simply needed to supply that same 5V to the daughter board and use it to supply the connection the compute stick. So, I desoldered the daughter board from the rest of the power supply, tested the voltage (it was outputting 5.14v – 5.22v to the daughter board), soldered on some new leads, and printed a very small box to house the card internally. At a cost of only 32g, I was able to power the stick internally and I gained two additional powered USB 3.0 ports inside the box which proved very helpful later in the design process.
The next bit of major circuit reworking dealt with the dew controller. I decided to use the excellent 4-channel dew heater control unit from Thousand Oaks Optical . But, obviously, I didn’t want the plastic housing or much else from the retail unit. I just wanted the control board.
To save weight and allow the board to sit as close as possible to the screen on the rear panel, I desoldered and removed the 4 output RCA jacks and the 12v lead and aux 12v RCA input. I also removed the primary choke and soldered it back on the rear of the board to lower the profile.
Next, I soldered new RCA output leads and 12v input leads directly to the back of the board and provided strain relief with hot glue.
One final bit of major electrical work before assembly. The QHY DC-201 has 5 status LEDs that provide information on voltage, TEC cooling, and camera fan.
I wanted to mount the DC-201 inside the Cube but I didn’t want to lose visibility to these status LEDs. So, I unscrewed the housing and removed the board. Using a straight edge, I carefully cut off the portion of the box with the status codes printed on it (I’ll use it later during assembly). Next, I VERY CAREFULLY desoldered the 5 LEDs being sure to label their position and polarity (LEDs are single direction electrical components). Using a very fine gauge solid strand wire, I re-soldered the LEDs to the board with 4″ long “extensions” protected in heat-shrink tubing.
Time to gather all the parts and clear the work bench for assembly!!!
First, front panel assembly.
Rear panel is next. The slots to the left and right of the screen allow me to attach a red overlay for imaging at a dark site with light policies.
The lower assembly begins by attaching the power distribution block and buck converter for the USB hub into their pre-printed spots on the lower box half.
Next, the USB hub is slipped into its custom fit wedge which features a lip to lock it in place. No glue or screws required.
Dropping the front panel in completes the front of the box.
Time to begin wiring all the switches, converter, regulator, and meter.
Adding in the dew controller power (later put on an internal 2.1mm plug so the rear panel could be easily removed for servicing the Cube). Also wiring in the 12v supply for the switch’s LEDs.
The Cube features 4 unswitched 12v aux jacks. These are a simple barrel connector soldered to leads and protected with heat-shrink tubing. I choose this particular connector because it came with a rubber cap to protect unused ports.
Adding in the aux jacks, the compute stick power supply, and internal USB 3.0 ports. Note the white USB-C cable to power the compute stick and the 3″ USB 3.0 cable connecting the hub to one of the internal USB 3.0 ports on the red power supply. The other 3.0 port is used for the touch screen (more below).
The compute stick is attached to its custom printed shelf with a simple compression latch. The shelf is designed to place the stick in the center of the cube for maximum ventilation.
The compute stick is attached to the screen via HDMI (for video) and USB 3.0 to mini for power and touch. While the USB cord is moderately flexible, most HDMI cables are not. Additionally, the connections are on the side of the screen with only a few millimeters of clearance. Luckily, there is a company that makes flat ribbon cable in various lengths between 5mm and 80mm with every type of termination you can imagine. You simply order the tips and ribbon and snap them together. I used a 20mm straight USB 3.0 to right angle USB mini and a 10mm straight female HDMI to right angle HDMI.
Dew controller connected and the lower half is complete!
The top half of the box contains the QHY DC-201 and the Losmandy dovetail adaptor.
First, the modified QHY DC-201 is attached to the roof of the top half of the box and the LEDs are inserted into their custom printed holes. The previously removed portion of the DC-201 is glued into an inset on the outside of the top of the box.
Next a custom 3D printed adaptor is attached to the top of the box to mate with a FarPoint dovetail clamp.
The QHY DC-201 is connected to the lower half power distribution with a 2.1mm plug and the two halves are screwed together.
Now, the only wire running off the telescope is a single 12v line. This can run from a battery or a well regulated power supply…I’m a big fan of BK Precision and use their model 1688B for my setup.
So far, this has really increased my imaging time. I can roll out of the garage, PoleMaster polar align, and be imaging in about 10 minutes. I can put the scope away in about 5 minutes. And I can run the whole setup from the comfort of my house with a remote desktop session to the compute stick.