Afternoon chaps

After finally getting myself a new job - I'm looking at getting my astro back on track. My initial thinking is to get the pier (NEQ6) fully automated with an embedded system such as raspberry pi, however I'm also thinking that odroid may be a better answer. I don't do the coding for that ATIK - I'll leave that to Peter and ATIK - however I may see if Peter is interested in an odroid deployment given it's additional processing power over the raspberry PI.

What I'd like is that the INDI system would operate headless from the embedded system connected directly to the mount and guider to minimise time, additionally I want to capture the *entire* guider image/star image to help with processing later offline (now I have my MBP GPU fixed!). This means I can WiFi into the embedded from a laptop and then if the laptop is put into hibernation, the system continues to collect data. I'm making the assumption that guiding is performed by kstars or the host platform rather than the embedded system.

Power isn't too much of a problem, although not having a million cables helps in the dark. The pier will have 12V power and possibly a ethernet connection or use a WiFi offload process for the images to a second machine for long exposures.

Has anyone tried the odroid?

Hi Nick, I actually use Odroid C1 for ARM development more than year and it is definitely a better answer. Peter

Excellent. I'll have a look into that.

Passive cooling on the C1 seems a better answer rather than the XU4 with a fan.

Have you tried a astrometry.net solve on it?

Thinking that putting the indices for astrometry on an SSD would make it very usable. Providing a capability of plate solving.

Not yet, but with SSD it is really quite fast compared to any RPi.

From the past - I know that the solver is was very single threaded, it may be possible to perform speculative searches in parallel. More of an astrometry.net issue. I've compiled this to a working state in the past from the sources.

Still trying to decide between a C2 and a XU4. I've read that the XU4 has some overheat throttling when all 8 cores are maxed however the more running on the card (linux, indi, kstars and astrometry.net) then it's likely to be better either using two C2 systems - one for everything else and one dedicated astrometry as the hammering it will give for bus and i/o during plate matches may impact the guiding). There's no reason for the plate solve to be sat as part of the realtime given the plate solve time.

Also anyone used the odroid show2 screen? I don't need anything explicitly flashy, just something that can show a cycling set of status pages with text next to the mount. A vt220 style screen may be very useful for that. Just thinking about the temperatures I've been out in (well below freezing).

Also - how much free space do you get on the eMMC with linux and indi?

Ok, I have pre-ordered a C2, 16GB eMMC and a show2 (small non-touchscreen display). I'll use a 9A 5V regulated supply connected to the 13.5V 9A PSU. That would be enough to power the 5V connections to two C2 boards, two 5V powered USB hubs with 8 connections (powering a 2.5" USB powered SSD, audrino focuser board, USB powered cameras etc). The rest of the kit is 12V (including the planned robot focuser stepper motor) and will take the mains powered 13.5V PSU output directly as it does now.

The plan is this:


Initially the single C2 will do everything, although I suspect that the astrometry.net would be more difficult and may need some external support. An update would be to provide a second C2 to act as a storage/astrometry.net server and connect the SSD to that.