Designing for and soldering a tiny BGA

t

I prefer to use tools I can trust.

So then you are agreeing with me. I don't think an integrated tool "solves" any problem. It just makes some operations a bit cheaper.

e

How does that work? If you want to change R9 to R1 and R1 to R3 and R3 to R9, where do you start? That is the real problem with renumbering is "proper" support. Just making both laytout and schematic update at the same time is the easy part.

That seems to be the netlist format that is commonly in use. I looked at some info on proposed standards for layout and netlist data. There seems to be a body that is trying to develop a standard, but even though most of the larger tool companies are on the committee, not many want to support what they have done.

Rick

me

? n

n)

e d

I use BullZip which will output PDF, PNG and JPEG among others.

Rick

I never trust CAD tools 100%. Because I've seen too much grief in that domain.

I does make hindsight changes much easier. But to me that advantage ain't large enough to warrant plunking down the equivalent of a Mercedes Benz for a new CAD suite. Plus I (usually) don't do layouts.

Renumbering is automatic, you don't need to do that. But:

You can decide way down in the layout phase "Oh, I better move this line one tap up the transformer because the paltry 3V headroom makes me lose sleep at night". I recently had to do that because a client's vendor began waffling about supply voltage at the last minute. Talking about a close call ...

Then you might find out that you really should have taken this new high faluting opamp that just came out yesterday. Unfortunately it has another package and pinout. No problem with an integrated tool. Just make one in the library, place it, flows right through.

If I did stuff like that to my PADS layouter he'd have a hissy fit.

Yeah, EDIF was just one big joke. At least some of the EDA execs might have gotten a few lobster dinners and cocktails out of it.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

We did as well. But in them's days those had round knurled metal holders. This sometimes caused them to roll off the drafting table and phsst ... tchok ... *OUCH* ... smack into the foot.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

ove

ld

l

I don't follow. Renumbering requires that you either use intermediate values of part numbers, or that you have to pay attention to the sequence of the renumbering. Unless your tools allow you to have more than one part with the same reference designator. My programs don't allow that since it is an error for most situations.

I just don't see where that is any issue at all. I did exactly that many times as I reconnected the FPGA I/O to optimize the layout. I just had both programs open at once and made the changes simultaneously in both programs.

Your PADS guy is a wuss. I have done wholesale replacements before and it is hardly any work at all other than the routing around the chip itself. I had a part on my last board that was a bit pricy. The only replacement was a smaller package. I designed in the smaller part (same pitch and pinout, but wider, SSOP instead of TSSOP) and later realized that I was now locked into that one. So I ended up changing the footprint so that either part would fit. For a layout guy to make a big deal of this is silly.

d e

The new format is based on XML. I get the impression that the format is good, but the users are just happy sticking with the current incompatible, problematic file formats and methods they currently use. I guess it comes down to what it would cost to implement vs. the benefit for them, not the customer.

Rick

Usually renumbering is fully automated and run twice: Once when the schematic is all done and ready for the design review. There you want R1 to be in the upper left corner. The 2nd time when the layout is done so R1 will be in the upper left corner of the board. No intermediate numbers needed on the part of the user, the software takes care of that.

If you absolutely wanted to do it by hand the old trick is to do a global change and add, say, 1000 to all ref designators. The old R1 would now be R1001, Q17 would be Q1017, and so on. Then have at it.

Well, I work in highly regulated industries such as medical. Once a netlist is generated that becomes a document in the design history process. Fudging with it in hindsight can raise the hackles of an auditor and that's never a good thing to happen.

Nah, he's a great guy. But as I said earlier this depends on what industry it's for. With an integrated system you have more freedom because less intermediate documentation is generated. It's like a CD you hold as a (non-Roth) IRA. If you cash it out at maturity and you carry the check to the bank with the next best IRA deal they'll send you a

1099. It's all fully legit but a lot more work than the "integrated solution" where an interbank transfer takes place and no 1099 is issued. Makes your tax returns a lot easier.

I've asked someone at a CAD company and they said that the willingness of the other manufacturers is the deal breaker. Or in sales speak, vendor lock-in.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

... snip ...

Well, it's been at least 20 years since I did a lot of medical software, and it was largely all done 25 years ago. Back then there were no such regulations to worry about, just the programmers own knowledge and conscience. My software was thoroughly checked out, but by me and my utilities. As an example, the only high level language I used was Pascal, and I had my own system, validated against the ISO standard. That was one reason for rejecting C at that time. I had control down to the last chip.

Another thing - the installations all carried CRC checksums of the binaries, which were checked at startup. At times startup was rare

- I had one system that ran for 3 years, until a mains power failure brought it down.

You can get a CP/M version of the Pascal software on my download/cpm page. Unfortunately much of the source got lost in a series of disk crashes and thefts of backup disks, otherwise I would have other versions out today.

--
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: 
            Try the download section.

Actually it isn't. Ok, I grew up in the days of DOS so maybe that's why it comes easier to me but except for a higher learning curve than OrCad on creating library parts I pretty much cracked open the Eagle package, installed it and was in business. The User language programming (ULP) is a very powerful tool that is missing from many other CAD tools.

But, for larger projects it's all water under the bridge because Cadsoft blew that one by not providing hierarchical design.

It isn't poor one. They all have their advantages and their drawbacks.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

I have never found automatic numbering to be suitable. I much prefer to have control over that myself. To do the "add 1000" think is also a PITA since you have to renumber everything twice. I have asked the creator of FreePCB to use an arbitrary ID number on each part rather than to identify them with the ref des. Then the ref des could be duplicated in the process of renumbering. A simple check for duplicates could be run after to make sure you didn't screw up.

I don't see how that applies. Your example is changing the netlist while in layout. How is that any different? If you have a locked net list, that should be *after* the PCB is in fab, not while it is in layout. I have never locked a design until after I completed it.

Your analogy is poor. I don't see the difference. In terms of engineering there is no difference.

Yep. Until the users demand that their designs be more portable, the vendors will just keep doing the same old, same old. Every time I work with a new fab house, I have to worry about communicating all the details that are outside of the documentation. It would be so much easier if it was all in one database that all the tools and equipment could talk to. But as far as I can see, in the CAD world, there is no desire or perhaps no ability to unify tools so that they work well together.

Rick

and

and

e e

r to use tool

,

ain

Maybe this isn't a good site gag, but de Nile isn't just a river in Egypt. If you think Eagle isn't a poor choice in layout software, you are living in denial. Just the problem of the Eagle virus along is enough to completely remove it from my list of potential tools.

No, you are right, Eagle isn't a poor tool. It is a HORRIBLE one.

Rick

Maybe this isn't a good site gag, but de Nile isn't just a river in Egypt. If you think Eagle isn't a poor choice in layout software, you are living in denial. Just the problem of the Eagle virus along is enough to completely remove it from my list of potential tools.

No, you are right, Eagle isn't a poor tool. It is a HORRIBLE one.

Rick

I have done some renumbering on FreePCB ... with a text editor. Even without looking up the format of the layout files, it is pretty easy to globally replace C11 with Cxx, then replace C12 by C11 and Cxx to C14. I sometimes change hole diameters too, much faster than clicking (what was it? 81 times for a 9 by 9 BGA).

I dump the files in CVS and get the revision number printed on the board automatically. Reassuring. There is no schematic to worry about, so there it ends. I'm a great fan of Free-PCB, but it may be just a matter of taste.

Groetjes Albert

--

--
Albert van der Horst, UTRECHT,THE NETHERLANDS
Economic growth -- like all pyramid schemes -- ultimately falters.
albert@spe&ar&c.xs4all.nl &=n http://home.hccnet.nl/a.w.m.van.der.horst

It's your preference not to do an automatic renumber but most engineers use it. This is why pretty much any CAD offers it.

I assume you are not working in the fields of medical, aerospace and such. There, the question is not when individual engineers think a design step should be locked but the question is when a standard operating procedure tells you when it will have to be locked.

See above.

I am afraid it's the first, there is no desire to unify file formats. I suspect the main reason is vendor lock-in. What can be a better deterrent from switching to the competition than a myriad of now unreadable design files?

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

I started out in medical 23 years ago and the regulatory world back then was _very_ different. But the grounds for change were laid pretty much at that time, mid-80s, by the Therac-25 accidents. Massive radiation overdosing, three directly related deaths. Those were caused by rather sloppy engineering, missing design review steps, probably a complete lack of a hazard analysis, and so on. That's when the FDA and others began cracking the whip.

Now Rick might not believe this but it's true, when auditors find holes in the design history or any other process then all the Klieg lights come on and they really start digging. One of our competitors (a very large ultrasound company) had chains and padlocks put on their building by the Federales. No kidding.

[...]

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

What Eagle virus? All I've heard so far was problems when people imported designs that had been made with hacked versions. And hacking software is a crime, no ifs, buts or whens.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

As I understand it...

Say you post a question about a tricky formula in Excel. I email you an example spreadsheet.

Two years later, Microsoft decide to make *your* spreadsheets unreadable because *I* used an unlicensed copy of Excel. Even when you prove to them your copy is licenced, they refuse to help you.

Totally unacceptable behaviour IMO.

--

John Devereux

Actually, there are plenty of ifs and buts, but most of those aren't applicable here. In any case, the problem affects users who have only ever used legitimate, unmodified versions, but were unlucky enough to have imported data from someone who used an unlicensed version (or who themselves did likewise, directly or indirectly; the behaviour is "viral").

It's quite possible that Eagle's "vigilantism" is itself a crime. There is a case (for which I can't find a citation right now) of an author who distributed a "trialware" application which encrypted the hard drive if it continued to be used after the end of the trial period. The author was convicted of computer misuse (aka "hacking").

This is in spite of the fact that it only affected people who had actually violated the licence terms. OTOH, Eagle's behaviour affects third parties who have never violated the license terms and had absolutely no way of knowing that a particular file would result in their data becoming "tainted".

It's been long ago that I looked into this, before buying an Eagle license. AFAIR it affects only the one file created with a rogue version which you import and possibly use as the basis for your project, not your other files. That doesn't sound viral to me. Viral means it would then go on and poison your other work that does not contain rogue project parts.

It's similar when someone buys stolen goods. Chances are the laws will find out and take it away from the buyer, regardless of the fact that the buyer had paid good money for it.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

R1

o

t.

Yes, that is what I said, I prefer not to automatically renumber components across a board without having any control. There are any number of features provided with tools that I don't use because they don't fit my method of working.

net

At this point, it appears that you are no longer reading what I am writing. So I'll just not reply to this.

Again, I can't reply to this.

o

In PCB layout as in many other areas, things are changing. The open source tools are becoming much more useful and widely accepted. My current customer actually has asked me why I am still using commercial tools when open source is available, such as schematic capture and PWB layout as well as HDL simulation and office tools. When my customers are asking me why I am being the dinosaur, I think that is a clear sign that I should pay more attention. I only wish I had stuck with Unix/Linux a long time ago. Then I would be over a big learning curve and already using many more of these tools.

Rick