Hi,
Having discovered the world of Raspberry Pi type processor boards, I tried to experiment with the different components I would like to use for a mobile and autonomous observatory.

With the new RPi2 it became possible to install also Kstars under Ubuntu but this device is pushed to the limit as far as the available resources are concerned.
I power my mobile setup with a 12V car battery with 74Ah capacity but use the own battery for my DSLR and the Tablet used as monitor and keyboard.
For the time being the following services need to be provided.
1. A Hotspot Wifi network allowing my different components to communicate over the tcp/ip stack (RPi2, tablet).
2. Time service synchronising the clocks of different components. A stratum 1 ntp time service using the gps time signal from the high performance satellie clocks.
3. Indiserver integrating all services.
4. Kstars with ekos for controlling the HEQ6-pro mount including alignment and guiding (using indi features, astrometry plate solving or phd2).

For the future I would like to add a filterwheel, an autofocusser and first of all a filserver sharing discspace with Camera's, logs, configs, observation schedules etc.
Via gphoto I would also like to power my dslr on the single battery based 12V power supply and remot control it.

I would like to ask members of the forum about their best experiences with the different processing boards available on the market. RPi2, Odroid, intel PC, .?.
Which hardware is according your experience best performing for running the services you need for observing in an autonomous and mobile way?
Many thanks in advance!!!
Stefan

What you really want to do is:
RPi( or RPi2) or Odroid running as a server, ie: controls your mount/dome/weather station etc....
Then connect to the server over the network from the comfort of your home/observatory room with your normal computer.
(If you really want to run all from the Rpi2, do a search on this forum and you will find some light Astro Distros)

Personally I have a single Odroid-C1 to run my entire observatory in Germany and I control it remotely from my home here in the UK.

Ok, so you run indiserver on the odroid and kstars locally in the UK?
Did I understood it well?

My specific issue is being mobile. I have no fixed observatory, so I move my equipment by car to a good observation spot.
In the field I do not have (always) internet access or power. The only thing I have is a satellite gps (hope in the future to use the Galileo clocks) signal.
Running the different services on different machines is of course possible but less efficient. In that way I could continue using RPi2 and may be an odroid for kstars but then I need probably a second battery.

stefan

Yes I have a fixed obs, remotely controlled.
check this thread out: www.indilib.org/forum/wish-list/904-rasp...-display-module.html it might give you a start for your project.

Hi, thanks a lot!!! This is exactly what I need. I'll have look and report back. regards, stefan.

Hello Stefan,
To answer your question on the asi forum.
I have no suggestions on the device (RPi2 or Odroid) to fit your needs. I have both odroid-C1 which is more like RPI2 and odroid-xU4.
The last one is really more powerful but is not really needed to handle indiserver. Furthermore the odroid-xU4 have only one USB2 port and I met some problems with the asi camera (only with it) because it only accepts to work on pure UBS2 port. On the opposite the odroid-C1 have 2 usb2 chip and 2 USB2 ports on each.
I have choose the xU4 because ubuntu 15.04 needed by the most recent versions of indi need it and it exists on this device.
Now I wonder if I can upgrade my odroid-C1 to ubuntu 15.04. This have lot of advantages: less consumption, less noise, and a 5V power feeding via USB.

What works for me now is to indiserver which serve asi camera for autoguiding, sbig camera for imaging and the mount. Some plugged on a battery, some on the main power (220V with AC/DC) and some others on battery. The main control unit is a laptop which runs Kstars and Ekos. Communication between the XU4 and the laptop is wifi.
I assume that the real problem is to configure properly the micro device to work as a hotspot and make its own wlan. (Was a challenge for me).

The goal for me now is to have all devices plugged on a battery. I will work on it during January.

In my opinion your needs are not really different. And I guess that RPi2 can do the job to run the indiserver.

Maybe I can help you too.

Patrick,
Thanks for your reply.
KStars on top of indiserver on an RPi2 is too much, but indeed, just running indiserver on it with different devices and KSTARS on a laptop might be the solution.

What is the autonomy of your laptop, if I may ask?
Is it allowing you to observe 4 to 8 hours? May be it is not necessary to have a working laptop with KStars and EKOS running for taking pictures during the night? Once the indiserver is launched and one uses the guiding module the system can run without any human intervention.
For tablets I see 8 to 10 hours autonomy but I am afraid that for laptops this is less and it is not always advertised.
I can of course buy a 12V to 220V AC converter but these things have an efficiency of less then 100% and with my battery of 74Ah I do not want to waste energy.
Stefan

I think the first issue one will run into when using the rpi, is the network connection. The network gadget on the pi is run thru the usb, and it's not a 'really cabable' network setup, cannot come anywhere near saturating a 100mbit link. That becomes a non issue tho if you are using wifi between the gadgets, because in general wifi is an order of magnitude slower than a wired link, no matter how much they hype various mbit numbers on the box advertising, you wont get that kind of thruput on a wifi link for a lot of technical reasons. This is the reason we ran gigabit out to the back lot where we plan to build the observatory. We will run the autoguiding software over the network, so full thruput capability is important to us. Current plan will likely have the cameras and mounts connected to a little atom based motherboard that has gigabit ethernet connections, so network thruput will not be an issue for us.

Years ago, my wife and I used to do the 'mobile astronomy' a lot, we had a camper van and would travel to star parties regularly, set up everything in remote locations then run it all for a few nights while camped there. Our setup was 'his and hers', consisted of two complete setups. One was an eq6 + c8, the other was an heq5 + 110mm refractor. Both were run from an atom based netbook, which we had bought the large 'extended life' batteries for. We had a 80ah deep cycle battery in the camper van, with an inverter.

Our experience was, we could go all night on a typical July or early August evening with that setup, but as the nights got longer in September, we would reach a point where we had to start the engine on the camper van because the battery would get low enough, the inverter would shut off. That was ok if we were out on our own, but it's forbidden at most of the larger star parties, no generators or vehicle motors running after 9pm. It was very important that we got the battery fully charged thru the day at those events. We usually ran a small generator for a few hours during the day to get the battery fully topped up if we didn't take the van for a drive that day. There were a couple mountain top events we went to every summer, and during the day we would drive down to a lake to go for a swim to cool off and clean up, and the camper van electrical system would do a good job of topping up the battery over that drive (an hour each way). The other trick to getting maximum life out of the battery, was to leave the netbooks plugged in for the first 2 or 3 hours and run them off of our main supply, but unplug them around midnight, then let them run from the internal battery till sunrise. The marketing literature bragged about 12 hour battery life from the big ones we bought, but, our experience was once all the astro stuff was running, they were good for somewhere between 4 and 6 hours, depending on how cold it was out. The colder it got, the less time we would get out of the batteries. Same issue for the deep cycle battery in the van, the colder it got, the less run we would get from it, and when ambient temps were around 0C, it only lasted half as long as running the same kit on a mid summer evening with temps in the 15 to 20C range overnight.

That was our original thought when we started going mobile, and we jumped thru all sorts of hoops to try conserve power and make things last. Then one evening when we had it all plugged in at home via wall sockets and 110->12V power supplies as necessary, my wife commented on 'how much better' things worked at home, vs in the field. We got thinking about it, and after looking at all the issues of trying to conserve power while on the road, we realized there is a much cheaper / simpler solution. Buy a bigger battery, then run everything via an inverter in the van that gives us a 110v plug, just like at home. Once we switched to that method, we never had a problem anymore, until batteries started to run down. Again, the simple / cheap solution was, buy a bigger battery. The camper van today has an 8D in it, which if memory serves correctly, is in the 160ah range.

Hi Gerry,
Thanks a lot for your extensive reply. Very useful!!!

About the bandwidth I agree but when I can keep traffic limited and data stored locally this is not a big issue. It depends of course on the work you want to perform.
I have run a vnc server on the headless RPi2 and connected remotely on my tablet in order to have a good resolution screen and a keyboard. I must say I was really delighted. Adding KSTars, the mount and a ZWO ASI CAM may become too much for the RPi2.

About the power supply, 75Ah is already a lot when running only on RPi2 and a mount. My battery is not a deep cycle battery so I wonder how long it will last. Because I am building up I have not had my first field experience yet. Logically I want to avoid transforming from one level to another and losing every time a few percent. Sticking on 12V supply is may be nice but today also an additional constraint complicating things.

I will carefully consider your experience.
Thanks a lot for the info.

Regards,

Stefan

Well, your logic is sound, except you are ignoring the largest power draw in the system, which is not the mount or the computer. It's the camera cooler. I've never done a full power budget analysis, because it gets variable depending on temperature, camera cooler works a lot harder as temps go up, but batteries produce less power as temps go down.

In the very early days, we used dslr cameras, and, to use those, you need to either run them from an external power source, or be prepared to change batteries regularly. When temps get down sub freezing, dslr needs fresh battery afer 3 or 4 exposures of 5 minutes, at least that was our experience.

Like I said initially, when we first started doing this, we focussed on 'energy efficient', and it was hugely problematic in the long run. Our final solution was to essentially abandon that idea completely, power everything from a single 110v plug per telescope, then just throw in bigger batteries to deal with it.

Hello Gerry and Stefan,

I agree with Gerry about the power sources. This is one of the main issue. I bought recently a Sbig 8300C which look to consume a lot of power to maintain a low temperature. I say it because I connected it to a 17Ah battery and see the voltage level decreasing rapidly.
I don't agree with Gerry with the bandwidth issue and agree with Stefan. If you do all the jobs locally, I mean taking pictures and storing them on a card the wifi bandwidth is enough to fulfil the needs. The ideal would be to have the autoguiding process embedded in the indiserver (maybe not the current situation, i don't think). So the traffic would be reduced to the minimum the exchanged informations between server and the monitor. I guess that processors on Rpi2 or Odroid are enough powerful to do a lot of job locally.
This is only my opinion but I have not enough experience to say that is truth

Regards

Patrick