On 11:27 AM 15/10/2002 +1000, Igor Gmitrovic said:
>All,
>
>I see there is a lot of diplomacy and politeness being flogged around. So, 
>before we all get civilised again, let me throw some more oil into the fire.
>
>In my opinion there are legitimate, logical and electrically correct 
>reasons to use 4-way junctions. E.g., if I wanted to represent the star 
>earth, I might even have 8-way junction.
>
>To me it doesn't matter how many connections there are on a junction if 
>they are correct. It is important that the schematic is readable and that 
>it produces a correct netlist and it correctly imports or exports to any 
>older or newer version I might use.
>
>The academic discussion conducted here does little to help solve the 
>problem of the incorrect import/export of junctions in Protel. There are 
>engineers who never held a soldering iron in their hand. So what? People 
>who only ever wanted to be sales engineers or managers don't need that 
>experience. It would definitely help them, but it is not necessary.
>
>This issue is fogging the real problem, and that is the incorrect handling 
>of the junction information in Protel. So lets concentrate on that. Please.

Here is what Nick Martin has had to say on this matter on the public DXP 
forum.  If you have a problem with it then I suggest you discuss it there 
as that is where the action on getting changes to DXP is happening.  There 
is little point in discussing the original bug as it will be fixed, in some 
fashion, shortly.  Better effort, I think, would be discuss the efficacy or 
otherwise of the fix.

Ian Wilson

The following is a copy of Nick Martin's reply to Holger Rusch's careful 
discussion on junctions four-way (or more) ties.
***************************************************
To: <[EMAIL PROTECTED]>
Subject: Re: [DXP] WARNING!!! Junctions at + points can disappear in DXP

Hi

This is a potential trap for 99SE users moving to DXP.

This has extensively discussed in the Pre-Release forum and a change has
been made to SP1 which removes this trap.
It only occurs on automatic placed junctions if you turn on the Optimize
Wires and Buses preference and when the set of conditions outlined below
happens.
The job of the wire optimizer to join pieces of colinear wire together so
this will remove wire end points.

So if you are using the practice of + points with junctions in schematic,
then make sure they are set to locked (manual) before turning on this
option.

In DXP, manual junctions are drawn with the centre 4-point handle,
autojunctions don't have this.
(Since Pins/Wires are also drawn with this handle on the hotspot it can
still be difficult to see.)

I posted the text below to the Pre-Release forum and I will post it here
also to help explain how to produce this problem and workaounds in the
current DXP release. As I mentioned before SP1 will alleviate this trap. (In
SP1 we find these special cases of AUTO-junctions on crossovers and replace
them with MANUAL junctions).

------------------------------------------
Intentional junction at + points should be OK since if they were manually
placed then they will be manual and so won't be touched by the auto-junction
system.

If you switch on "Optimize Wires and Buses" then it will change any broken
line (with pieces) into a single wire object. So as far as I can see the
sequence to generate a problem would be:

- In SE (or DXP with Optimize Wires off), place a wire.
- Place another wire that butts up to this to form a tee (autojunction
should appear).
- Place another wire that converts the tee into a + point (autojunction
stays).
- If the design is in SE format then import into DXP - no changes should
occur in junctions.
- In DXP, turn on Optimize Wires (its off by default), and move or place a
wire or bus.
- The system will optimize the wires and the special case mentioned above
will create two crossing wires and the autojunction will be removed. No
changes will occur to manually placed junctions.

The autojunction system looks for points that have three points (exactly
three) intersecting, or a single point intersecting on a horizontal or
vertical wire (but not on the end). Any autojunctions that don't meet this
criteria will be removed and any points that match this criteria will have a
junction added.

The easiest workaround is to leave Optimize Wires off if the Schematic has
any questionable design issues off this kind.
Another way is to do a global edit and change all the existing junctions to
manual (makes later editing harder).

------------------------------------------

Best Regards,

Nick




----- Original Message -----
From: "Holger Rusch" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Monday, October 14, 2002 5:01 PM
Subject: RE: [DXP] WARNING!!! Junctions at + points can disappear in DXP


Hi,

 > I agree, the tools should not rip them up, but I still think
 > they are a really bad design practice.

Hmm, doing it that way for 5 years now. Are you constanly fax your sheet
to somebody? Its 2002. The sheets arent archived on print. Use small
style for junctions and they are big bobs.

 > Yes, they are shown differently in DXP now, but they were not in 99SE.

They dont show different in DXP now. The default in DXP is to place
locked junctions. If you switch the locked option off the look similar.

The worst is to import an SE99 design with + junctions into DXP. => It
gets ripped up. No notice, no nothing.

--
Holger Rusch

____________________________________________________________
Considered Solutions Pty Ltd     mailto:[EMAIL PROTECTED]
ABN: 96 088 410 002
5 The Crescent
CHATSWOOD   2067
Ph: +61 2 9411 4248   Fax: +61 2 9411 4249

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* 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/proteledaforum@techservinc.com
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to