<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>
Date   : Fri, 08 Oct 1993 02:54:48 +1300
From   : David Andrew Sainty <David.Sainty@...>
Subject: Re: introduction

From: J Day <jeday@...>
Date: Thu, 7 Oct 93 14:01:06 BST

>> > University of Edinburgh.  I use three BBC Master Systems to control
>> > experimental apparatus via the user port.  I am currently trying to
>> > build up a system which records events via an interface and writes
>> > datafiles to disc in real-time.  My ideal solution would be to obtain
>> > some Winchester Hard disc drives for the Masters so data collection
>> > would be a lot easier.  I would welcome any help on how to obtain such a
>> > device.  I have been told that it could be possible to network the
>> > machines and use an old BBC B as a fileserver to write to and from disc.
>> > 
>> > Could anyone help?
>> 
>> Can you be more specific on the sampling speed and volumes you're
>> looking at here?
>> 
> Yes.  I currently run three separate Master Systems which collect files
> about 150k per day.  Each file is built up quite slowly, but at maximal
> speed would need to write 2 lines of data per second.  If the three
> computers were linked to another to act as a fileserver then the disc on
> that machine would need to have three separate files open each with a
> capacity of 150k ish.

A master can store 640k (I think?) on one 80 track disc. It can certainly
keep up with the data rate you require. Your easiest solution would be to
drop two of the masters and just use one. This means slightly more complex
hardware, but saves you networking, or having to use several disc drives,
and leaves you with a couple of masters free to use Tequilacomm on. :-)

Only other problem is what the hardware is actually doing, because whilst
the disc drive is saving data the speed you can look at the user port will
be reduced. No problem if the rate you are talking about is reasonably
consistant, but if you are talking about very high speed bursts of data
you may have a problem.

Hmm, here's a way around that... Use two masters, one set up as above,
but buffering the data and sending is at a slow but consistant rate
throught the serial port to the other machine that then saves to disc.
Linking two machines with the serial port is cheap, and with the server
dealing with the disc drive, all the possibly high volume work (watching the
user port and writing to the drive) are isolated to one machine each.

But I think you can do it all with one machine...
<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>