Alt-BEAM Archive
Message #00209
To: Jean auBois aubois@trail.com, beam@corp.sgi.com
From: "Ben Hitchcock" beh01@uow.edu.au
Date: Sat, 06 Feb 1999 11:03:53 +0000
Subject: [alt-beam] Re: Nv ring Synchronizer (was: BEAM clock)
Hang on, there's a way to do it.
If you got a reliable 1 hour (or 2 hour, or 4 hour, depending on how many
hour neurons you wanted) clock working, then synchronising the minute part
is a snap.
Just use an open-ended network: Once the pulse gets to the end of the
chain, it falls off the end: ie. don't connect the last neuron back to the
first one.
So each time the hour part fires, it starts a new process in the minute
part.
And if you made the hour part open ended as well, then the clock alarm would
start the hour process off as well....
Easy!
Now where did I put those 3600 uF caps and 1 meg resistors....
Ben
----------
>From: Jean auBois
>To: beam@corp.sgi.com
>Subject: Nv ring Synchronizer (was: BEAM clock)
>Date: Fri, Feb 5, 1999, 21:55
>
>At 12:41 PM 2/5/99 , dennison wrote:
>>Synchronization? Seems to me I posted a not to long ago about something for
>>just that.... :-)
>
>Well, sort of... what you describe was more of a device that -indicates-
>coincidence rather than causes synchronization. In other words, pulses
>have to come along from both chains to the NAND gate at the same time to
>change the output. That change in output could, indeed, create closely
>sync'd pulses back into the chains or some other chain. What I'm talking
>about (and perhaps this isn't even the right way to do things) is more like
>one loop _causing_ a change in the other loop (moving to the next hour
>indicator, for example).
>
>Like Ben wrote, accuracy would almost certainly be extremely poor -- that
>was, in fact, what I thought the biggest challenge would be. The hour loop
>would most likely slip with regard to the minute loop and as a result the
>pulses would almost never arrive at the NAND gate at the same time.
>Besides which, what is required is that the minute part of the design
>causes a -different- state change in the hour part every time it cycles.
>Merely injecting a pulse from the minute loop into the hour loop doesn't
>help 'cuz the hour loop would just become saturated (very slowly).
>
>Actually, I'm afraid that Nv/Nu/Suspended technology can't do this without
>referring back to the discretized world of registers and counters. For
>example, the "on the hour" pulse from the minute loop could be used to
>-clock- a single process around a loop of Nv latches (they have two
>resistors connected to the bias point: one to Vcc and the other to ground.)
> However, this is essentially a shift register and doesn't benefit to speak
>of from BEAM/VSPANS.
>
>By the way: I'm not building such a thing at the moment simply because I
>still don't know how to do it.
>
>
>Zoz
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com
Home