Re: gEDA-user: M68K bus and transmission line ringing

[Gene Heskett <gene.heskett@xxxxxxxxxxx>]

On Friday 19 October 2007, Michael Sokolov wrote:
>Hello fellow gEDA/pcb users,
>
>I know there are a few old-timers on this list (by that I mean hardware
>engineers of old school), and my question is directed to those.  Would
>anyone here happen to experience designing a system with the good old
>Motorola MC68302?  (It's an old-school telecom processor, if I'm not
>mistaken it's an early 1990s chip, but it most perfectly fits the late
>1980s computing and communications environment, the era of MicroVAXen.
>Has the most classic original 68000 core inside.)
>
>I'm using MC68302 on my SDSL board, and I'm now in the process of
>putting the finishing touches on this design before sending it off to a
>layout contractor.  Here is my area of concern: my current schematics
>(see http://ifctfvax.Harhan.ORG/OpenSDSL/OSDCU/) have the M68K bus
>interconnecting the MC68302, RAM, flash, the SDSL transceiver's
>microprocessor control port and the FPGA.  I have pull-up resistors on
>the bidirectional control signals (AS, UDS, LDS, DTACK).  But I don't
>have any series resistors yet, and I wonder if I should add some, and if
>so, where?
>
>I've been told that when a sufficiently fast-switching signal is driven
>onto a sufficiently long net, one has to worry about this signal ringing
>due to transmission line effects.  That of course raises the question of
>just how fast it needs to be and just how long do the traces have to be
>for the issue to become a concern.  I've been told that my MC68302
>running at 16 to 25 MHz is fast enough, and that traces running half-way
>across my 130x165 mm board would be long enough that I do need to worry
>about this stuff.  I've been told that the solution is to insert series
>resistors of ~30 Ohm into the nets close to the source.
>
>Here are my questions which I hope might be answered by someone who has
>some experience with MC68302 or any M68K-based design:
>
>* Does the MC68302 in fact produce slew rates fast enough on the M68K
>  bus signals it drives (address bus, data bus on writes, control
>  signals) for the designer to be worried about transmission lines
>  ringing?
>
>* As I've mentioned before, my design will be laid out on a 130x165 mm
>  PCB.  I can probably get the core microprocessor subsystem (MC68302,
>  RAM and flash) fairly close together in one corner, but I also need
>  the bus to go to the SDSL transceiver chip.  The latter is a mixed
>  signal IC and its physical placement is rather inflexible as it's a
>  critical component standing on the boundary between the digital and
>  analog sections of the board.  The latter requirement would probably
>  mean that my M68K bus traces *will* run half-way across my PCB.  Is
>  running half-way across a 130x165 mm PCB long enough for me to be
>  worried?
>
>* If the answers to the previous two questions are affirmative (i.e.,
>  that I do have a transmission line ringing issue which needs to be
>  addressed), what's the proper way to address it?  Would series
>  resistors take care of it?  If so, exactly where should I put them?
>  On the address bus?  On the data bus?  On the control signals?
>
>* I've been told that the series resistors I'm talking about go right
>  after the source.  But where is the source on a bidirectional
>  multipoint net?  Take the data bus for example: it's bidirectional and
>  goes to multiple peripherals.  Series resistors between the MC68302
>  and the main bus nets would take care of writes, but what about reads?
>
>I hope someone can give me some insight, preferably backed by some
>actual experience with M68K designs.
>
That I can't give, but what I can suggest is to find a decent tome on 
transmission line theory just to give you some background.  Also, any 
multi-point scheme is probably going to a cut & cry solution.  Even the act 
of probing a trace to see what the signal looks like can change things enough 
that a working circuit will fail, or a failing circuit will work.

Circuit board trace widths & spacings to other traces all become data to be 
optimized in situations like that.

The usage of a src termination r is often a grand compromise, and may have 
more to do with the FCC's radiation of a computing device pass/fail tests 
than in actual circuit performance.  More than one consumer device has been 
rendered very undependable by those requirements.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Revenge is a form of nostalgia.


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user