Eagle library editor

part

levels become

out of

That's

I meant that my signature is the only one that authorizes an ECO. The author signs it too, and the document control manager signs it when she releases it, usually the same day I authorize it.

John

Reply to
John Larkin
Loading thread data ...

then to

be

numbers for

is that

didn't

ECO

changes.

We have a stock number (7 digits, like a phone number) for every part, and that's what's on a BOM. For each stock number, there's a list of authorized manufacturers and their part numbers, which is how Purchasing knows what to buy.

It's up to Engineering to make sure the purchasable parts are equivalent. We have a few cases where, say, Fairchild part ABCD is one of several acceptable buys for our part xxx-1234 but the only allowed buy for xxx-1235.

Do you use manufacturers part numbers on your BOMs, with no second sources?

John

Reply to
John Larkin

NO! A physical PCB assembly born rev A is forever rev A. It's etched and silkscreened on the bare board and stays that way. Rev B is a new document release, from which new boards are etched. No physical board ever changes rev letter.

A box that contains multiple boards could be rev B, but have an assortment of PCBs and subassemblies inside of various revs. All that is controlled by its BOM. We roll the rev letter of such complex assemblies if the change could affect a customer, namely if the box FFF changes.

John

Reply to
John Larkin

for

that

We do have an informal system for making engineering-level boards that will never see production, test circuits and fixtures for example. They get a "99/Z" series prefix and their files are dumped into a "Protos" folder on a server, and purchasing picks up the gerbers from there. There's no formal release or document control, and often no BOM. But these are never shipped to customers. Well, almost never, but they are understood to be test boards.

Here's one:

ftp://jjlarkin.lmi.net/Z250A.jpg

John

Reply to
John Larkin

writing home

design is

defined by

schematic he

ECOed

separate

BOMs,

There

the

calls

When you want to rev the bare PCB in a CAD program, how do you know which schematic was the old one, which is to be the new one, and what do you call the new schematic?

PADS creates a change file (which they call a forward ECO) by comparing schematic rev C (the old one) to schematic rev D (the revised one), which we apply to PCB rev C to create pcb rev D.

How would you do it, when the schematic gets revved independent of the PCB? Sounds like a management nightmare.

John

Reply to
John Larkin

part

levels become

least

out of

ECO615".

That's

my

and

are

be

any

intended

order a

far

was

to be

No, in that case you release an ECO to change it back (what you have to do anyway). If you guessed that a deviation was the way to go and have to change it, just release the same information in an ECO. It's not a nightmare at all. Just good business sense.

But you've said more than once that the only name needed on the ECO is yours. That sounds like a pretty aggressive top-down business to me.

But manufacturing doesn't have any "you gotta be kidding" input on the matter. Neither does PC or purchasing.

Sure, but the buy-in has to be corporate wide.

Depends on what they are. Actually, I'm too new in the company to know the process yet. I've only logged into Agile a couple of times and can't find a damned thing. All of the specs (what I've been looking for) are in there somewhere but I'll be damned if I can figure out where. There isn't even any tree to browse around in. :-( The place has been pretty empty for the last couple of weeks (use-em-or-lose-em) so there hasn't been anyone to ask, either.

Reply to
krw

part

levels become

least

out of

ECO615".

That's

my

and

Ok, still pretty authoritarian, but at least it's just a JH (the rest of the signatures are there). Most release the ECO when all of the relevant people have signed. No "over-seer" needed.

Reply to
krw

then to

be

numbers for

is that

didn't

ECO

changes.

That's true for any part, though. One battle we fought constantly was faster op-amps would detect the 2.4GHz carrier. Some would, some not. Since the radios were TDMA at 100Hz or 200Hz, audio was sometimes an "issue".

Reply to
krw

That's the stuff. Do you remember what you paid for PADS? Was that a per-module pricing? Was/is there a yearly "license" that has to be renewed?

Reply to
JeffM

writing home

design is

defined by

he

ECOed

separate

BOMs,

There

the

In the olden days of paper schematics and tape-on-mylar PCB artwork, the schematic was just a reference drawing: it didn't control configuration. With CAD programs, that's no longer the case. The schematic is the ultimate control drawing, because it creates the netlist and BOM that propagate all through the system. A change to a PCB involves processing two schematic revs, the old one and the new one, and the PCB winds up hard conforming to the new one.

A lot of document control is hearsay and tradition. It doesn't always make sense.

It's hell-to-pay if your schematic and PCB stop having the verifyably-same net list. It makes sense to us to keep their rev letter the same.

John

Reply to
John Larkin

then to

the

be

history.

numbers for

is that

didn't

ECO

changes.

The parts master entry often contains second sources. But only if, say, you have a MMBT3904 from ON Semi, ST, Fairchild, NXP and so on. It does not extend to a FFF-compatible transistor with a different number, from a different process. That will get a new part number. In order to release that one into an assembly the whole assembly needs an ECO. Which (usually) bumps the rev level, especially if the old part has become unobtanium.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

then to

the

be

history.

numbers for

is that

didn't

ECO

changes.

That is one of the very little-known secrets. "Why the #%^! does 1.8GHz stuff get into this LF amp?" ... "Because the first BE junction rectifies it" ... "Oh s..t!"

Often if the 1/f knee wasn't so important a switch to a CMOS opamp "fixed" this. But that requires an ECO :-)

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

That's definitely a no-no in medical devices. The document rev level for a circuit board assy must always match the assy rev level. If it doesn't then Monsieur l'Inspecteur will write a ticket.

I meant the board FFF change. An IC gets replace by another IC.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

to

numbers for

that

changes.

But the one in your example sez REV A :-)

Nice. J15 and J19 must be massive connectors.

On my boards there are always two part numbers. One for the bare board and the other the assembly number. So far I've never had a client that does it differently. In the old days the bare board P/N was in copper but with some of the more fancy solder mask colors that can be hard to see. So now it's mostly both in silk screen. Personally I prefer yellow, like on your board.

Another thing we used to do in the 90's was a staggered layer indicator at one edge. That would immediately show a layer screw-up at fab.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

writing home

chip design is

defined by

schematic he

PCB

be

wasn't ECOed

separate

in BOMs,

were

the

There

using the

calls

the

The schematic and BOM can reference the board part number, including its rev level. Mostly in the form of an extra sheet which also has the mechanical hardware on there, such as heat sinks, shields, screws, card throwers and so on.

In the layout program it's really easy, the P/N plus rev level is on the silk screen file :-)

Not a nightmare at all. The board is a part of the schematic, just like all the transistors are. Because you need one bare board to make this assembly. If you need a new layout then it'll be a P/N with a new rev level.

--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

writing home

design is

defined by

he

ECOed

separate

BOMs,

There

the

Not necessarily. Yes, they are linked. But you can keep the rev A bare board yet have a new schematic. For example, R92 might go from 12.1k to

14.3k but the layout doesn't have to change for that. Now the bare board rev level remains A but the assy goes to rev B.
--
Regards, Joerg

http://www.analogconsultants.com/
Reply to
Joerg

then to

be

numbers for

is that

didn't

ECO

changes.

That's a minor point. We release letter revs without an ECO; naturally new rev BOMs are released at the same time. We literally interpret ECO as Engineering Change Order, namely a directive to take exception to a released document.

It's all logged and visible.

Where is that kept? Who can mark it up?

The ECO form has fields that say "Effectivity" and "Reason for Change" and "Changes" and "Actions." People are supposed to fill them in properly. Engineers aren't always super articulate in writing, which is one reason I read them all.

When we roll a PCB design (schematic and PCB together!) we include a README file that says what was actually done and why.

John

Reply to
John Larkin

then to

the

Must be

history.

numbers for

problem is that

didn't

the ECO

proto

X1

changes.

I tried finding a CMOS opamp that would drive 150ohm headsets hard enough (needed a 12V rail and differential drive). Space was a consideration, too. The driver was a MSOP and there wasn't much room for anything else. There was one Intersil part but I never got to try it. The other guy found a Chinese garage that had apparently bought up the old Moto masks and was making them crappy enough. ;-)

There were plenty of ECOs on that board (up to 'M', IIRC).

Reply to
krw

Release a rev without an ECO? You are out of control.

If there is no ECO how is it logged?

Since it isn't in the ECO system, it's kept by the engineers. There is also a "problem report", which is tracked by the system. Anyone can enter a problem report and these get "answered" before the next ECO.

Note that this "working copy" isn't used for the next rev, rather a "clean" copy from the last lettered ECO is used. The "working copy" is what we generally use around the lab, and like the paper version had our "scribbling" on it.

They should be right before the ECO is signed off.

Again, that's in the ECO. There is also a note on the first page of the schematic but that's not the controlled source. The ECO is the last word.

Reply to
krw

writing home

design is

defined by

schematic he

ECOed

separate

BOMs,

There

the

calls

But the board rev stays the same. The assembly rev changes. The BOM controls the assembly. The schematic controls the raw board. Again, several products can be made from the same raw board.

A board part number of 1234567A is a raw board and its schematic is forever linked with it. The first release of the BOM will be called 1234567AA (the second 'A' marked in sharpie on the final assembly) and the second 1234567AB, and so on. ECO 1234567AA will have the schematic, Gerbers, and everything in it. Under normal circumstances (no hacks) the ECO for 1234567AB will just be a BOM and assembly instructions. The BOM calls out board #1234567A.

Reply to
krw

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.