At 05:53 PM 9/6/01 -0500, Jeff Stout wrote:
>I really hesitate to follow up this thread because nobody is
>going to change anybody else's mind.
I am not writing to change people's mind, at least not the mind of people
who are set in their ways. Rather, I'm writing to examine the issues, and,
as I have noted elsewhere, there is a rather pressing need to examine this
particular one.
> The only reason I am
>responding is to give Brad a little support for his POV vs. the
>overwhelming decient.
>
>I too use 123 as opposed to EBC. I think it has three big advantages.
>
>1. I only need one PCB footprint for each footprint type.
It is agreed that this is, in one sense, an advantage. However, it is
counterbalanced by the need to have umpteen different schematic symbols,
and, typically, the difference between these symbols is not apparent on the
schematic and it is therefore relatively difficult to check -- unless the
numbers are shown. And didn't at least one of the proponents of using
numbering state that he suppresses the numbers? (As do many of us when
using numbered pins on transistors and FETs. Perhaps that is not a good idea.)
In order to have only *one* numbered footprint, one must settle on a
standard. And no matter what standard we accept, we will cause confusion
when the manufacturer's numbering on the data sheet is different from that
standard. I think most of us with substantial design experience have seen
the errors that this causes. It would be wonderful if we always checked
everything thoroughly and never made mistakes, it might seem, but if we
were to consider how much effort that might take, it might not be so
wonderful. Time is money. I'd say that anything we can do to make errors
less likely, particularly if it is a simple and cheap thing to do, is worth
doing.
> Take
>a TO-220 for example, hole it upright and pin 1 is to the left,
>pin 2 is in the center, and pin 3 on the right. A TO-92 same thing.
Yes, especially with the TO-220 part. I've never used anything other than
numbers for a TO-220. There remains the problem of what to call the
mounting tab, but I've usually called it "4", or, sometimes, "MH."
I've no problem with TO-92s, per se, but when one converts the same design
to surface mount and uses a SOT-23 for the same functional part, often with
*almost* the same part number, things get dicier. Using letters will force
one to change the footprint, and if the footprints are pinout specific, one
will be forced to deliberately select one of the various possibilities.
This forces a consideration of the problem, and thus it is *much* less
likely that an error will be made. Otherwise -- and I have seen this happen
more than once -- one simply assumes that the numbering will be the same,
and one only changes from TO-92 in the footprint field on the schematic to
SOT-23. Mistake, and one that often survives moderately careful checking.
Of course, some of us have learned how easy it is to make this mistake, and
will review these assignments. Usually. ....
>Yea, things got more complicate for SOT-23's and such, but I put
>a crude drawing of oddball footprints on the schematic so
>anybody can find pin 1, 2, or 3. etc on my PCB's.
Not a bad idea; such an idea would be just as helpful in an environment
where numbers are used. Remember, we are talking about letters vs. numbers
as they appear in the net list and in the PCB footprint. But, having been a
tech, I would much prefer that the diagram show me BCE than 123! It is one
less step in the mental process....
2. I don't have to make a million redundent footprints. Imagine
>you need a voltage regulator, or a reset chip in a TO-92
>package, I don't have to make yet another TO-92 footprint;
>the same 123 footprint works here too.
Ah, but you do need to make another schematic symbol. I find it easier to
modify footprints than to modify schematic symbols, but that may also
reflect the environment in which I work, where I don't necessarily have
write permission for the schematic. Generally, schematics are drawn by
engineers who may or may not do their own pcbs. Obviously, the situation is
different for many of us. Some of us are engineers who do their own
schematics and design their own boards. I might note that normally this
would be a very small company, struggling to make ends meet, because
electronics engineering time is typically more valuable than printed
circuit design time.
Anyway, using numbers essentially implies that it is better to take time to
make different symbols than to take time make different footprints. And I
think that is backwards.
If an engineer has used letters, all he needs to do for me is to give me
the part number. From that, I get the pinout and assign the appropriate
footprint. But if he has used numbers, I have a more complex process to
follow; I'll often have to resort to looking at the net list to determine
what net a particular pin number belongs to, and then I can assign the
footprint. Remember, I cannot assume that the engineer has followed some
presumed numbering standard, and often he hasn't. Sometimes he has used the
manufacturer's pinout -- and who can blame him?
>3. Imagine your a technician looking at the schematic. If you see
>a BJT, the emitter, base, and collector are obvious without the
>EBC labels. Now you look at the PCB and which pin is which is
>not obvious at all.
And it is not obvious, period, unless one knows that a standard numbering
is used. The technician is going to need that information. Mr. Stout says
that he puts package drawings on the schematic. That's one way to provide
the information. But it would be even more direct to put the functional
names or abbreviations on that package drawing and skip the numbers
completely. With this, one does not need to know the numbering at all, it
does not matter what standard the designer has used.
>However, if you use 123 system the tech looks at the transistor,
>sees the base, emitter, and collector and the pin numbers. He then
>goes to the PCB, finds the Ref.Des. and can find the base, emitter,
>and collector (all the footprints are the same remember).
But he has to know that, and he has to know which standard has been used.
That is a no-brainer for a TO-92 but just about completely arbitrary for a
SOT-23. If the tech works for you, you can inform him. But most of us sell
products that are used by other people, and sometimes they, too, need to
look at the circuits!
I realize I'm not going to change anybody's mine, but I thought
>I give Brad some support.
>
>Give it up Brad.
I want to emphasize that Mr. Velander is one of the most valuable
contributors to this list. My comments here are not against *him*, but
rather against an idea that he has promoted, an idea I think mistaken if it
is taken as a general recommendation.
>Jeff Stout
>P.S. I'm not that pure either. I use AK for diodes... bad, bad, bad.'
The situation with diodes and SOT-23 packages gets even worse.... If one
has used AK, whenever one changes the package for the part, one must change
the footprint. That, I think, is what we would expect with *any* package
change.
Sometimes the footprint is topologically the same, sometimes it is
different, but the rule for the designer to follow is simple and
intuitively obvious. But the schematic, from an electronics point of view,
has not changed at all.
And this is an advantage. To change packages, one changes the part number
-- if that is carried in the schematic -- and one picks the footprint that
matches that part number.
If numbers were used, the schematic symbol would have to be changed *and*
the assigned footprint must be changed -- in some cases. Not only is it an
additional step but it is more error-prone.
I'd estimate that the use of numbers for SOT-23s and the like costs Protel
users and their companies or clients a million dollars a year, multiplying
my experience by an estimate of the number of designers.
[EMAIL PROTECTED]
Abdulrahman Lomax
Easthampton, Massachusetts USA
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To leave this list visit:
* http://www.techservinc.com/protelusers/leave.html
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
*
* Forum Guidelines Rules:
* http://www.techservinc.com/protelusers/forumrules.html
*
* Browse or Search previous postings:
* http://www.mail-archive.com/[email protected]
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *