At 02:54 PM 9/8/01 +0200, Soren Juel Andersen wrote:
>If you support the EBC, AK system, how many pin does a package need to have
>before you switch to

I think that was previously answered, but I will answer it again. When 
there is a standard numbering for a part, one uses numbering.

>What about 3 pin regulators, FET's  with 2 gates and so on.

Yes, letters for those, except, again, where there is a standard numbering. 
I used to think that TO-92 packages were numbered in a standard way, for 
example, 1, 2, 3, from left to right with the flat down, as seen from above 
(opposite the pin view). Then I saw one which had been numbered 
differently. But that is such a wild exception that I'd have no problem 
supporting numbering for TO-92 packages. On the other hand, being able to 
change from a TO-92 package to a different package, say a SOT-23, without 
changing the schematic symbol, would be an advantage to using letters.

The basic point is that using letters forces one to investigate the 
package. If using letters is the standard, then one will have different 
packages for different pinouts. I've been doing that for years.

>Nobody would dream of using labels on a 7400, you do not a hundred
>different DIP14 footprint.

This reductio ad absurdem argument would only apply to an attempt to apply 
the proposal in an absurd way. DIP14s have a standard numbering -- the 
exceptions stand out like a sore thumb. So we would continue to use 
numbering, and this applies to most parts with more than a few pins.

The dangerous situation is where a part has a standard package, but the 
silicon has been connected in various ways by the manufacturers, and the 
pins are numbered differently by various manufacturers. To repeat, where 
there is a standard numbering, numbers are better, likewise if there are 
many pins. (But with parts such as pin grid arrays, letter-number 
combinations may be best, though they bring their own problems. Really, in 
this situation, letters are being used as numbers, i.e., as symbols where 
the order is significant.)

>During service or debugging you will have to have the schematics anyhow.
>Is it not a tradeoff between the schematic designer (who could be a little
>lazy, and are using the supplied schematic package as is, e.g. the symbols
>in the standard library NPN,PNP ) and the PCB designer, who
>should have the image of the many transistors.

There are tradeoffs, certainly, but the use of numbers, if the numbers are 
suppressed on the schematic -- as was indicated -- conveys *no* information 
unless the technician knows how the numbers have been assigned on the part. 
In a tight company environment, there might be a standard; even there it is 
one more thing for a new hire to learn.

Technicians in the field do not normally have the CAD database, so the pin 
names in the net list and in the PCB footprint are irrelevant. Generally, 
he or she will see a symbol on the schematic and will be faced with a 
physical part on the board. To relate the two a drawing of the part or 
knowledge of the same will be needed.

If one really wants to help the technician, one will put functional labels 
next to the pads, i.e., B,C,E or whatever. Putting numbers would convey 
same information but in a roundabout way, and only if the numbers are shown 
on the schematic (or, again, are already known to the tech). And if you are 
going to put functional labels, why not have the same labels as pad names?

I'm arguing for the use of letters in standard library parts under the 
following conditions:

(1) There is no universally -- or nearly universally -- accepted numbering 
of the part.
(2) There are relatively few pins. (There are varied numberings of QFPs, 
for example, but I would still use numbers because functional names would 
be totally unwieldy, and, as noted by Mr. Andersen, one would need to make 
a huge number of footprints.

To my view, the only reasonable alternative would be to establish a 
standard numbering. And establishing standards like that would be nearly 
impossible outside a single-company environment. Single-company 
environments, if highly customized, can really be a disaster when the 
company needs to expand or get help.

Why do I think this important? Because we have a library committee and we 
will need to decide about pinouts. I assume that this discussion will be 
considered by the committee. (Membership in the library committee is 
presently open and all are welcome. If you are interested, join 
[EMAIL PROTECTED] Presumably, the committee will make 
recommendations which will be published for for review and comment before 
being accepted.)

Abdulrahman Lomax
Easthampton, Massachusetts USA

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
* Contact the list manager:
* Forum Guidelines Rules:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to