[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: starship-design: HIGHLY OPTIMIZED TOLERANCE
John Bean wrote:
> I don't know if I agree with the make-everything-myself idea when it comes
> to really complicated processes that are really expensive such as chip
> making. At some point you need to repair/replace the "table top wafer fab"
> and the technology to do that would have to be carried. Then that
> technology needs to be repaired so you need the technology to do that. This
> goes on forever with each iteration adding weight, complexity and cost.
That is only true, with a evolving system. The Critical components
would be fixed in design. Barring the latest PC clone I can see only
50 - 60 different kinds of semi-conductor components. A air circulating
does not need a 1 GHz 64 bit cpu to control it.
> Also making chips is not as simple as a wafer lab according to the guys who
> are making custom chips for me. The process is big by nature, requires lots
> of energy, lots of chemicals and consumables, very skilled people and
> ultra-high tech ultra expensive equipment. These guys snicker or drool
> whenever you talk about a small or inexpensive or easy prototype machine
> like you are discussing.
That is true, but semi-custom chips require less processing, at say a
loss in speed and processor size. High density programmable logic,
that they be used wisely and you don't have problems with bit rot in the
that could really cut down the number of spares needed.
Right now I am working with a semi-custom design package, ( free on
with rather slow logic ( 1 ns gate + 1 ns/pf ) to design the ttl logic
processor I always wanted to build but never could make up my mind on
the instruction set. Since I started programing computers on the PDP-8,
a classic micro-controler with 4 k-words of core memory, I have this
for bloated code. Anybody need a 24 bit processor chip ... coming real
soon. I am sure that even the old 8 could handle control of the fusion
reactor or plasma drive.The PDP-8S ( slow ) was a neat all transistor
( 1001 of the suckers) computer.
You only need computing power for graphics,games
and multi-media. People stuff, not ship running stuff.
> Even if you could create this machine and have it weigh 100 pounds with 100
> pounds of support stuff, excluding consumables, and 100 pounds of spare
> parts and equipment it becomes really heavy and would cost millions of
> dollars. I don't know how to figure out the weight of the person to operate
> this or the supplies he needs. This also ignores the toxicity of some of
> the processes.
Very heavy!? you just said 300 lbs. Most likely 3,000 lbs and the same
for support stuff. Note this would be at the trailing edge of technology
as most of the design stuff would be simple devices.
> A better suggestion may be to design a "generic" logic module using modern
> technology. I'm not suggesting a few gates on a board as in the 80s I am
> suggesting a 32 Bit ARM processor, a DSP, 160k of program flash, 8Mbits of
> data flash, 12k of RAM, a USB port, an IRDA port, some A/D channels, some
> D/A channels, somer general purpose IO pins. This could be built on a 2x3
> board that uses very low current (50ma with both CPU and DSP cooking at
> 22/80 mips) at 3.0 volts using available (off the shelf) technology for $20
> a board in medium volumes at about 2 oz a board.
Eprom,Dynamic ram,Flash,Cmos all have a limited life span (10
by cosmic radiation and just is over kill for many things.. Low density
bipolar fuse based logic looks to be more rugged in a space application
> In place of your 300 lbs of machine you could carry 2400 spares for $48000.
> Considering a desk-top fab will probably cost Millions, which excludes the
> packaging and testing equipment also required.
Funny last time I looked at star-ship prices that was in the travelers
role-playing game and they where not cheap. For a fleet of 12 ships
design was only 10%. I place this as a .1% cost modification.
> The real difference is that you have to get used to the concept that in
> some things, such as modern electronics, you are often better off to throw
> a bunch of parts at a simple task than to make something custom.
Is not that the same thing, custom chips only are the bunch of parts
all on one chip rather than a PC board.
> If you need to interface a full keyboard use one of these boards and also
> use one to interface to a single switch. The idea is to have so many of
> them that it just doesn't matter. If the initial design used this approach
> there would be thousands of these boards in a starship spread through
> critical (life-support) and non-crital (personal entertainment) systems.
> Take some multiplier of these as spares and you would be covered for many
Take keyboards - I want a just a keyboard , not a key board with
trackball, infrared remote and windows keys just to punch up the
commands to "leave orbit"
or "hyper speed 3.623".
> Since many of the boards will be used for different types of tasks a
> particular failure on one part of the board would not prevent the board
> from being used in another location. For instance if an AD failed switch
> that board to some task that doesn't use the ADs and take the one from the
> task that didn't use the AD. This provides some redundancy without even
> reaching into the spares.
> This approach can be applied to other technology issues also. For instance
> only use two kinds of displays, a big one and a small one, and make sure
> the big one can do the job of the small one. Then carry spares, mostly of
> the big ones.
That is the best way to design it.
> Another way to extend the mission duration is to send supply ships ahead,
> or send them faster from behind with some replacement stuff. Since "stuff"
> doesn't need gravity or environmental controls or oxygen or food or water
> it can be moved much faster and much cheaper than we can move people.
> Although re-supply may be distasteful to "pure" starship travel the
> reallity is that even here on earth it was, and is, considered normal.
> Aircraft carriers and submarines and space stations do it, the only
> difference is distance.
Or move more at a slower speed, only "living" things need to be moved
> Considering FTL isn't going to happen soon, if ever, it is probably safe to
> assume that we would go pretty slow in a colony ship. If a supply ship can
> go 1.5 times faster at 8 months we send one with the first resupply
> reaching us at the 1 year point. Assuming engines will keep getting better
> perhaps the next supply ship is launched at 18 months and reaches us in 2
> years. Radio will still be effective so we can provide them with a shopping
> list even if it is taking weeks or months for it to reach earth.
> Personally I don't think we will be going to reach the stars soon. However,
> I think starting to travel is critical to our survival just like setting
> out in doughouts was to most early sea fareing nations. I think that our
> travelers, now in space for 15 years, will be stuned when some kid pulls up
> in his dads Mach5 with his FTL Quantum drive with the Window's logo on it
> and says "haven't you heard, Bill Gates bought NASA, now everyone has an
> Intel FTL drive in their family car. Other than having to turn it back on
> every hour and the occasional warping into a star it works great"
"We do not inherit our time on this planet from our parents...
We borrow it from our children."
The Lagging edge of technology: