Rich <someone@somewhere.com> writes:

In semi-infinite wisdom Louis Scheffer answered:

"Rob Dekker" <rob@verific.com> writes:

"Louis Scheffer" <lou@cadence.com> wrote in message
news:4095e404$1@news.cadence.com...

Andrew Nowicki <andrew@nospam.com> writes:

[...]  As a first guess, cover the planetary zones
of the nearest million stars or so, with beams that are on
all the time  [Think of the beams as looking like a pincushion].

This is something we can do with existing technology for about
$200M, for a beam bright enough that we ourselves can detect it.
And this cost will come down further as a consequence of Moore's
law.

How did you get to just $200M for a system which cover 1M stars;
semi-continuously ?

Here's how I got that number.  Assume you build a large phased array consisting of
many small (10mw) transmitters.  To be detectable by ourselves, you need about a
10^12 watt EIRP.  1M of these implies a total EIRP of 10^18 watts (it turns out with
phased arrays the total EIRP can be split up with almost no penalty).  This means
you need 10^10, or 10B transmitters.  We can build today a 10mw transmitter plus
antenna for about $0.20 (actually we can build a chip with 100 of these + a PC board
with 100 antennas for about $20, but for our purposes this is the same.)

This gives a total cost of $2B for the transmitters.

Think you'll need much more than that, power for starters. 

Power supplies are quite cheap at this level.  At 50% efficiency, you need about
20 mw per transmitter.  PC power supplies are about $0.10 per watt. So this only
adds $0.002 per transmitter, or about 1%.

And unless I'm greatly confused, you'll need an active interface to
point the arrays. I'm not sure how these arrays are controlled, but
I'd imagine that you'd need to least a serial connection to each board
for control. And what kind of a system can direct 100 million boards,
simultaneously? I don't think anything like this even exists. 

This is a real problem, but this is not as hard as it looks.  
Each chip can compute its own phase evolution, reducing the 
communications and control to something practical.  I've
written this up in some detail, see

http://www.lscheffer.com/papers.html

and look for 'A High Power, Low Cost Microwave Transmitter for Deep Space
Applications'.

We'll need lots of supporting infrastructure, and it seems to me that
that'll cost more than the antenna boards.

Basically. think of the floor of a large shopping mall, but with PC
boards instead of tiles. Concrete slab is not that expensive, and all
you really need is stability, and not precision.

However, you are ordering 100 million identical PC boards and chips,
so you should get a substantial volume discount.

I agree with you there. :-)

How much?  I'm guessing a factor of 10 since it would be reasonable to
set up a dedicated production line for these volumes.

And how long will they last exposed to the elements? How about rain?
Dew? Insects? Animals? Birds? Bird droppings?

These are reasonable questions, but I suspect they also have reasonable
engineering answers.  Solar cells, for example, are warrantied by their
makers for 25 years, and they are exposed to all these things.

Come to think of it you'd probably need an enviornmental impact
study as well.

Almost surely, for something this size.  But the power density, 16 watts
per square meter, is only 1/60 of sunshine.

How much power do you need for the transmitters ? 

About 200MW is required for 10^10 10mw transmitters at 50% efficiency.

Have you factored in the cost of power generation?

If you agree to buy power, someone else will be happy to pay for generation
if you agree to buy energy long term.

However, you could also make the array
solar powered to cut the operating expense

Solar cells are expensive however, up front costs would be huge,
regardless of operating costs. 

Solar cells are are indeed expensive, but might pay off.  A 30 W panel
costs about $200 today, so this would double the costs.

And I'm not aware of any large solar facilities that show low operating 

costs. And the same issues with exposure apply.

We would not need power conversion hardware, which is a big part of current
wind/solar costs.  Solar cells have of course dealt with exposure 
issues for years, and come with a 25 year warranty.  This is one of 
the reasons I think the PC board exposure problem is solvable.

Maybe, but the power feeds would be interesting, 140MW total to antennas
2.5 cm apart. 

This is relatively easy.  Perhaps one PC power supply (320W) for every 10
square meters.  Perhaps 100 of these fed from a 32KW transformer, 100 of
these from a 3.2 MW transformer (building size), and there are 80 of
these.  All these are stock components.

I expect that there would be significant cooling issues also.

No, we only need to get rid of 1.5 watts per chip, or about 24 watts
per square meter.  No forced cooling is needed.

And when an element fails, how are you gonna get to any of them for
replacement? You could not walk between the rows. 

Every few tiles, you have blank row just for the purpose of walking
through for service.  Every 50 meters or so, you have a road's worth
of blank rows, so you can drive your service truck.

And just how sensitive >is the array to antenna failure(s)?

Not very.  If 1% of the modules are dead, the EIRP is reduced by about
2%, and presuming constant replacement of dead modules you should never
have this many dead at one time.  You do need to make sure command and 
control of the remaining modules arenot affected, though.  This is 
covered in the paper.

I think you are underestimating the actual costs by at least an order of
magnitude, perhaps more. YMMV.

This is conceivable, but it's still cheaper than the big dish approach
for any reasonable expense model.

     Lou Scheffer