Roll-off roof fail-safe closing - how about gravity?

[happz]

Me and my friends are thinking about building a remote, fully automated robotic observatory, and the bit I'm most worried about is how to close the roof when power supply crashes, controlling Raspberry Pi dies and so on, and avoid crushing the telescope by sliding roof at the same time. The location is cca. 2 hours by car from my place, therefore I'd like to be sure it's possible to safely close the roof even in such case. We were thinking about using gravity to slide the roof back to the "closed" position. Opening the roof by an engine, lowering the remaining side - just like on the "Doghouse" picture but with the roof resting on inclined rails - and hold both parts in their "open" positions by some sort of electromagnetic lock, which, when not powered anymore (or told to do so by computers) will disengage and let roof slide down to its "closed" position. Seems to be doable, it has some advantages and some disadvantages as well (e.g. it's necessary to stop the roof somehow, right?  ).

What caught my attention is the fact that I haven't found a single observatory in the wild using this approach to fail-safe closing, and I wonder way. Has anyone any experience with this solution? Since probably nobody constructs their observatory's roof this way, there must be some crucial error I cannot spot

[CraigT82]

I'm no engineer but would it be much easier to just install an uninterruptible power supply?

[happz]

18 hours ago, CraigT82 said:

I'm no engineer but would it be much easier to just install an uninterruptible power supply?

I'm no engineer either, and UPS would help a lot, no doubt about that. But at this point I'm quite curious about the solution I proposed above and whether it's viable or not, other people's experience and comments.

[pete_l]

How would you ensure the roof closes if (for example) ice forms on the rails, or the sliding part freezes in the open position? Or if the wooden framework warps in the sun?

The lengths you would go to depend a lot on the value of the equipment you want to protect.

 

I have used RPis quite a lot. I would not trust one to always work when it is needed. Personally I would want a second system available independently. But then, 2 computers is as bad as two clocks: if they disagree, which one is correct ?

[Adam J]

I would use a batter backup system that automatically closes the roof if the main power is cut. 

[Skipper Billy]

If the power fails how will you park the scope before the roof crashes into it?

[wimvb]

In a horizontal orientation, the roof's motor only has to overcome friction. If a roof closes by gravity, the motors will have to overcome friction plus gravity to open it. This generally costs more, since you need a stronger motor.

The usual solution is a UPS, or a large enough battery to park the scope and close the roof. Probably a cloud sensor and/or weather station, so that you can monitor the situation.

The budget solution is a good neighbour at the remote site, who can help in case of an emergency.

[ollypenrice]

You've identified the key problem. I'm not sure about the solution. If you did use gravity you could do it with flat rails and just have the roof lift a weight on a cable as it opened. I think the idea is worth thinking about.

However, the first thing to avoid like the plague is having a roof that can collide with the scope under any circumstances. We have a 'no collision' policy on our pair of robotic sheds. This is not a problem with scopes of even quite large size though the biggest amateur scopes might be hard to house in this way. The trick is to make the upper sides roll with the roof so that the scope can see over the fixed dwarf walls.

This is our small single scope shed housing a 10 inch RC. Note the self closing end flap on the south - a simple but effective solution. This would easily scale up to house a 14 inch or larger. (By putting a weight at the front of the OTA it can be mounted further down the saddle plate needing less roof height.)

So the best and simplest solution to the collision problem is to eliminate it at source.

Power outage can be mitigated by a UPS for the computer and a battery for the roof motor.

But that still leaves a computer crash, which happened to us once. The roof opened in the rain. (This is why I only host on the clear understanding that I know of no way to elimniate all predictable risk and that the IT management of the sheds is the responsibility of the owners. I can see no way round this.) Optics are remarkably tolerant, electronics not so! The obvious thing is to put as much of the electronics as possible under cover within the shed. That's easy. With a bit of thought the cameras, also, can be protected from the worst of it by simple shielding. 

Would it be possible to wire in a rain sensor which would over-rule any instruction (including no instructions) from the computer and trigger a closure in the rain?

I honestly think that, in the end, there just is an element of risk. But when one of sheds did open in the rain there was no long term damage to any sheilded components and, after that, all the owners added to their sheilding!

Olly

[pete_l]

9 hours ago, wimvb said:

The budget solution is a good neighbour at the remote site, who can help in case of an emergency.

I think that is an absolute must-have as  a last resort. Even if all they can do is send you a photo of what has happened for you to assess whether you need to drive out to rescue anything.

[happz]

21 hours ago, Skipper Billy said:

If the power fails how will you park the scope before the roof crashes into it?

In my case, the roof could not collide with the scope, it'd be rolling off with part of sides as well.

[happz]

23 hours ago, pete_l said:

How would you ensure the roof closes if (for example) ice forms on the rails, or the sliding part freezes in the open position? Or if the wooden framework warps in the sun?

 

This can happen even with UPS, right? I guess that especially warped framework is probably nothing backup computer/power system deal with.

23 hours ago, pete_l said:

I have used RPis quite a lot. I would not trust one to always work when it is needed. Personally I would want a second system available independently. But then, 2 computers is as bad as two clocks: if they disagree, which one is correct ?

Yes, that worries me as well. I plan to use at least 2 RPis, maybe some external watchdog as well to at least let me know both died, but I suppose when they disagree, I'll shut everything down and deal with the broken one later.

[happz]

 

3 hours ago, ollypenrice said:

This is our small single scope shed housing a 10 inch RC. Note the self closing end flap on the south - a simple but effective solution. This would easily scale up to house a 14 inch or larger. (By putting a weight at the front of the OTA it can be mounted further down the saddle plate needing less roof height.)

 

Oh, I love it! I'm very good at what I do, dealing with software and bugs, but when it comes to hardware... I couldn't invent anything like this

The no-collision policy is definitely the way I'd like to follow, rolling off the roof *and* part of side walls. I was hoping, in the worst case of power loss or computer crashes, I'd be able to let gravity do its job without any risk of collision between the roof and the scope.

 

[wimvb]

30 minutes ago, happz said:

This can happen even with UPS, right? I guess that especially warped framework is probably nothing backup computer/power system deal with.

Yes, that worries me as well. I plan to use at least 2 RPis, maybe some external watchdog as well to at least let me know both died, but I suppose when they disagree, I'll shut everything down and deal with the broken one later.

 

On 2017-06-20 at 13:42, pete_l said:

I have used RPis quite a lot. I would not trust one to always work when it is needed. Personally I would want a second system available independently. But then, 2 computers is as bad as two clocks: if they disagree, which one is correct ?

The space shuttle had an odd number of control computers (3 I think), just to avoid the risk of a 'hung jury'.

Unfortunately, if you use quad core computers, you can never reach an odd number. But you could have twelve jurors.

And now the important question: should I add a big  to this?

 

Have you checked indilib.org? The software is used for robotic observatories. It does a graceful exit (= park scope and close roof) automatically when connection to the client is lost.