RE: [N8VEM-S100:665] 6502 board resistor networks

["John Monahan" <mon...@vitasoft.org>]

Not to beat this thing to death but further for my comments on my iPhone
yesterday (complete with its mis-spells etc):-

In general there are three types of pull-up resistors on S-100 boards. In
all cases the decision is between power consumption and signal shape/slew.
If there are just a few pull-ups on a board one usually uses a low value
like 1K.  However a number of boards have quite a few, thus the use of
resistor network chips.  Let's look at each case.

1.  Tying an unused TTL input high.  
Usually 1 K is used - though you can definitely use lower values.  If one of
a dual input (e.g. 70LS00) sometimes both inputs are tied together. Sucks a
little more power but for LS ok, probably not for 74Sxx or 74Hxx.   If the
input is shared with say 5 or more other TTL inputs fan-out/loading starts
to become an issue. Signals start to slew. Use a separate 1K pull-up rather
than tie inputs together.  If you are board power utilization concerned, you
can use higher values for a single unused input at least up to 4.6K.  At
these values susceptibility to board noise can be a worry at high speeds.

2  Tying open collector outputs high. First must be done.  1K is most
typically used for the 1-10MHz ranges we typically use.  You can go higher,
however in critical areas (for example clock inputs to LSI chips where
signal slew is important) at least 1K is desirable, some cases 500 Ohms.
However in general for fast critical signals you should not use OC driven
signals anyway.

2. Outputs to S-100 bus lines. This is where it gets complicated. Much has
been written on this. Where the bus is terminated is a big factor.  1K is
usual for OC lines BUT that assumes only one board is pulling the OC line up
on the bus.  If you have multiple boards (or front panel) with the same line
being pulled up the OC signal can run into problems. In these cases 2k, 4.7K
is often used.  That is why on our 8086 board we have these pull-up
resisters with a jumper option.   For bus lines the higher the signal
frequency the lower the pull-up signal should be to get sharp rise and fall
times. 

Seeing some comments from time to time here I think people worry too much
about resistor values. First most resistors are +- 10%.  Only if you really
want to tweak the board for absolute max speeds does the value come into
critical play.   In the boards Andrew and I have done, if there are
absolutely critical values we will mention them as such. Offhand the only
ones I can remember are a few on the ZFDC board around the WD2983 chip.

Hope this helps
John





John Monahan Ph.D
mon...@vitasoft.org

-----Original Message-----
From: n8vem...@googlegroups.com [mailto:n8vem...@googlegroups.com] On
Behalf Of Neil Breeden
Sent: Thursday, January 12, 2012 11:14 AM
To: n8vem...@googlegroups.com
Subject: Re: [N8VEM-S100:665] 6502 board resistor networks

This is the kind of discussion that should be captured, cleaned up and
placed in the wiki for the newbies / those without EE knowledge. Part of the
N8 effort is to help people gain knowledge and grow skills.

-N

On Jan 12, 2012, at 10:26 AM, yoda wrote:

> I agree with your statements - I think there can be some relaxation of 
> some things.  Some parts are difficult to come by now and I would hate 
> to discourage new people from participating because of that - so there 
> has to be a balance.  Not all participants are electrical engineer 
> types so it can be frustrating to them not knowing where they can 
> substitute.
> 
> On Jan 12, 10:59 am, "j....@cimmeri.com" <j....@cimmeri.com> wrote:
>> Seems like there's different, perhaps competing, objectives.
>> 
>> One objective is to faithfully recreate legacy designs to build and 
>> experience legacy systems that might not be easily done any other way
>> (such as finding a particular real legacy card on eBay).   My own
>> interest is in legacy systems, but for example, finding a real legacy 
>> 6502-only board has been impossible for me, so the n8vem 6502 board 
>> is the next thing.. at least it uses a legacy design and parts.  
>> Really, what parts cannot actually be found in some way, unless they 
>> were custom like PALs?
>> 
>> Another objective is that of evolving S-100 into more modern designs 
>> and operating systems beyond, say, 1986, to beyond where it had gone 
>> during non-obsolescence eg. PC compatibility with MSDOS (aside from the
>> CompuPro effort), or Linux.   If the intent is to produce a set of whole
>> new designs to delve into these later systems, then these designs could
>> easily have consistent standards across the board.   This objective is
>> of no interest to me, but I'm sure many would get a kick of out 
>> running newer systemologies on the old S-100 standard.
>> 
>> Though these objectives clash if not clearly defined and left 
>> confused, they can in fact be pursued simultaneously... with a clear 
>> distinction made between a legacy offering and a new offering.
>> 
>> - John Singleton
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> lynchaj wrote:
>>> Hi Dave,
>> 
>>> Thanks!  I like the idea of consistent design guides however most of 
>>> our S-100 boards are legacy designs.  Most are wholly or partially 
>>> legacy designs from a multitude of sources.  For instance the S-100
>>> 6502 CPU board is a PCB implementation of Rich Leary's home brew 
>>> board used with permission.  I tried to be as close to his original 
>>> design as possible to improve the chances of a working PCB.  Similar 
>>> for the
>>> S-100 68K CPU board, I got permission from Alan Wilcox to reuse his 
>>> design.  John has a variety of home brew S-100 boards and design 
>>> elements from a mixed bag of sources. With so many different 
>>> designers it is no wonder we are seeing so much variation.  Every 
>>> board has its own story!
>> 
>>> We do try to have some consistency across boards and reuse design 
>>> elements when possible but there is still a lot of variation.  Many 
>>> of the boards come with their own unique legacy and are particular 
>>> to their original designers.  As the designs mature, I think we can 
>>> make them more consistent through respins and tweaks but it will 
>>> take experimentation and just plain field experience to find out 
>>> what changes we can make without breaking the board.  My personal 
>>> approach is to be conservative and faithfully replicate the original 
>>> design as closely as possible.  I generally don't stray away from 
>>> the original drawings especially on the initial version.  There are 
>>> just so many variables in conversion from a home brew wire wrap 
>>> design to a PCB that broader design consistency tends to take lower 
>>> priority over basic functionality and reliability.
>> 
>>> That being said, I think there is a lot of room for improvement as 
>>> the board designs mature.  Consistency is something we can "grow 
>>> into" or at least reduce the wild variations to something more
manageable.
>>> Thanks and have a nice day!
>> 
>>> Andrew Lynch
>> 
>>> On Jan 11, 10:12 pm, yoda <y....@r2d2.org> wrote:
>> 
>>>> Hi Andrew
>> 
>>>> Would it be possible to have some design rules in general.  I have 
>>>> seen a lot of these boards use parts that are not easily obtainable 
>>>> which suggests these are copies of old boards without thought.  If 
>>>> they are supposed to be pull-up resistors then in general I would 
>>>> expect 1K or 4.7K be specified as they are pretty standard.  I 
>>>> checked Jameco, Digikey and Mouser and they don't have 1.3 K.  I 
>>>> know experienced people can interpret schematics but it tends to 
>>>> discourage new people into the hobby that don't have that 
>>>> experience.  Also it would be nice to do some standardization of 
>>>> buss interface.  I see this board uses ls541's where most other 
>>>> boards use ls373's so one has to "stock" many more parts to
participate.
>> 
>>>> Just a thought
>> 
>>>> Dave