From: crunch@netcom.com (John Draper)
Date: Mon, 9 Nov 92 13:39:59 PST
To: cypherpunks@toad.com
Subject: Analysis of cost to produce random serail generator.
Message-ID: <9211092136.AA25999@netcom2.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain


At Hackers 8.0,  we had a discussion about the feasability of someone
building a random data generator.    I volunteered to look into this
with the hopes of making a cheap and inexpensive device that has
2 DB-9 connectors (Serial through box) to be used to generate random
serial data at a particular baud rate.

Today,  I called Newbridge Micro,  they faxed me the data sheets,  and
gave me the number for a local rep in San Jose.   (408) 258-3600,
but I was appalled at the price.    They wanted $52.50 each in
100 quantity.   So it is clear that if I use this chip,  which really
is a hybred,   I cannot charge $50 each.    So we have to decide
what to do,   or how much more it will cost.

After examining the data sheet,  the chip is probed by a pulse,  and
a bit comes out with an equal probability of a "1" or a "0".   Sorta
like a coin toss.  

I think that the perifery cost of the other electronics,   such as
shift registers,  or UART which would be necessary to clock the bits
into an 8 bit register,  etc,  would cost about $6 - $8 in cost,  then
I will have to fab the boards.   I have friend at Douglas Electronics
who can give me a good price.   It would cost $150 setup charge for the
PC boards and $2 per board. 

1 ea  RBG1210           $52.50
1 ea  PC board		 $2.00
various chips		 $8.00
Setup 			 $1.50   (1/100th setup charge)

Total parts		$63.50  (100 quantity) 

Cost (4 X parts)       $254.00   

So,  it is clear that the cost is rather high,   and not everyone can
afford this.     But if you think that people will want to buy it at
that price,    I can go ahead and do it,    but the cost to me would
be about $250 to build just one of them,  and I have someone who can
loan me the money to make a prototype.   This also includes design
time and use of an electronics lab to test and get it running.

I wouldn't be able to borrow $6350 to build 100 of them,  and I think
I can talk the rep into letting me purchase them in smaller quantity,
so I can build them on demand.    I'm just not so sure that people will
want to pay $254 for one of these.    So please lets discuss this,  
among our group to determine if this price is reasonable.

Although,   it IS possible to use a cesium noise source (Don't know
the cost of that)  and perhaps I can cut that price down by about
a half or a third,  but the design time would be much increased,
and perhaps there would be twice as much electronics,  and perhaps
the posibility that the randomness won't be as good.

John D.