The (one-and-only) version that generated the gerbers. And the parts list.
Most of these operations in our case do not require a schematic. Only one might be board repair, but maybe not even that these days. A
*printed* schematic does not typically contain enough information in any case. It is the BOM that specifies the company part number which allows the part to be looked up or replaced with the correct type etc.
Cute - in fact our system can do that and also with (optionally) "invisible" text+data points. But I never thought of doing little pictures. I used to do it with those data points but tend not to bother these days, it's easier to add the parts in after the BOM export. Subsequent revisions include all previous parts automatically unless explicitly changed.
That wasn't the point. Of course our modules also have a BOM each. The difference versus yours is that they also have their own personal schematic each. It gets release together with them BOM and all the other stuff needed for production.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I know, that's why I ask how you do what you wrote above: How does the "and it's noted on the schematic" happen if you design a new filter but use the same PCB? But you gave the answer further below, the correct values aren't really always on your schematics I assume.
So the new values are not noted on the schematic, it seems. That would be just great for the poor guy who has to repair stuff.
Ok, that sort of answers it. That would be unacceptable to pretty much all my clients, I'd say. Monsieur l'Inspecteur would also wrinkle his forehead about that.
Yes, absolutamente.
No, each gets its own fresh part number. Because it _is_ a functionally different part, it is not FFF-interchangeable with any of the others. So it'll be something like this:
7683-0001 Rev A
7683-0002 Rev A
7683-0003 Rev A ... and so on.
Also for us:
7683-0001 Rev B
7683-0002 Rev B
7683-0003 Rev B ... and so on.
In this day and age it is easy to push such information without felling a tree first. Signing multiple ECOs where each has the same info changed is easy as well. It was a bit more work in the late 80's but even then, it had to be done.
In production usually not. In engineering, yes. In the same way that you cannot have out-of-calibration equipment in production but in engineering you can. From a regulatory POV those are entirely different countries.
Our worlds might be a bit different in that respect. In medical, if someone makes a mistake based on the scribbles by a guy from the night shift, people can die.
That's because you never get audited, and the tie breaker rule probably isn't even written down anywhere. You just can't do stuff like that in med devices. I have witnessed many situations where that would have be really great. QC guys already left for the weekend and such. But we could not do it because it would have violated a procedure.
It's like flying. If breaker #43 keeps popping and you are sitting in your own airplane you could make a conscious decision that it only feeds the cabin lights and take off anyhow. In a commercial airliner you look up the procedure and if that says it must be fixed by a certified mechanic you shut down the engines and call dispatch.
There's a note on each sheet "Engineering Orders May Apply." That means that the values you see may not be the values that are on every version. The schematic does not define an assembly; a BOM does.
Out test and engineering folks understand our documentation system. If they question a part value, they pull up the BOM. I don't mind if they make pencil notes on their schematics. Those notes don't control configuration.
We'd tell M. l'I to get lost.
Looks like our dash numbers. The difference is that we wouldn't issue
That isn't very helpful to the guy having to diagnose a big module.
With many of my clients that is perfectly ok for engineering folks. But test folks penciling in notes? Absolutamente not.
In med devices that would be the same as telling the CHP officer that just pulled you over to get lost. Pretty soon there'll be 2-3 more cruisers with flashing lights.
An hour ago it was 40. That's inflation ! :-)
Well, we have to, or would anyhow because it makes a lot of business sense. The exceptions are very small modules with a limited number of versions. There, the schematic may contain a table spelling out the component values for each version.
Having component values only in the BOM isn't going to fly in most cases.
[...]
Lucky you. Many of my clients do not have that choice. Well, unless they elect to give up production which none of them would do.
As I said the repair situation is the only time where some schematic may be useful. In fact we don't make much where a schematic would be useful. But the "gerber schematic" combined with a parts list gives all the information. If we made something like a radio or SMPS where end users might need the schematic then perhaps we would generate one as part of a service manual or some such thing.
What's a gerber schematic? Gerbers are simply layout file sets, there is no schematic. Plus some parts may not be stuffed and there are no values at all visible in the Gerbers.
Well, most of the designs I deal with are quite complex. The set of schematics is a sizeable stash of sheets and the BOM is huge. It's like trying to repair your LCD TV set without a schematic :-)
Yes sorry that was unclear. I meant the schematic used to create the netlist of the artwork file, which was used to create the gerbers. It is always very close to the stuffed board, and fine for "understanding" the circuit. But it is possible something like a resistor value was changed in the BOM or a section of a multichannel design not fitted etc. It is definitive as far as the netlist is concerned, and is used as the basis of future revisions in conjunction with any existing ECOs/component changes.
It's not that our boards are not complex enough to need a schematic. But our customers are not electronics companies, we don't *want* them trying to repair a 0.4mmm pitch QFP with their soldering iron (that they also likely use for soldering copper pipes). The only people repairing the boards are our own staff who have the parts lists and access to company part stores.
Isn't it a potential problem for someone to repair a board using a part they guess is right purely from the schematic information? Or does absolutely everything that is relevant to a part have to appear on the schematic? Enough to let a random "electrician" order a replacement from Farnell?
You wouldn't allow a test person to write anything on their paper copy of a schematic? Not a sketch of a waveform? Not a typical voltage? Not a note about a dipswitch address setting?
It is very hard and a waste of technician time if they have to thumb through the BOM for every part value that has or might have changed over the years. Not a very efficient process, IMHO.
Same with my clients. Access to parts lists and parts stores does not help during the diagnosis. You need current and up to date schematics. Not the old set from 1998 when the layout was done.
They don't. They use the schematic to _diagnose_ a problem. That takes the lion's share of time, usually. Once they identify a part that needs replacing they key in "C17", the software looks it up in the BOM and automatically pops over to the MRP system, tells them it's P/N xxxx-xxxx, hit print, a move ticket is printed, they give that to the stock room clerk who then hands them the part. The department gets cross-charged $0.17 for that. Sometimes the move ticket is electronic.
In med or aero there is no guessing, this all goes by procedure.
No, that's in the BOM which gets accessed from the company-furnished computation machine. We did progress a bit from the days of horse and buggy :-)
Via the computer, yes. But they normally do not order anything, that is done by the purchasing department. Which can escalate the order if this is an urgent issue.
Yikes. If someone here wants a part, for whatever reason, they walk into the stock area and take it. You're supposed to fill out a "swipe" sheet, but we don't bother for, say, one resistor off a reel.
My clients generally don't either. For example, at EndoSonics our production manager and I instituted a Kanban style "pull method" so they could have hand stock of anything they wanted. Of course, they did not do this for big fat 20lbs power supplies or $300 FPGAs because we'd get dinged with excess inventory if they did.
Walking into the stock areas was also quite restricted. We had areas with sterilized goods and one lone mishap could have forced us to scrap tens of thousands of Dollars worth of finished goods there.
You might think this is all silly but let me show you with just one example:
One of the things I design a lot is patient interfaces. These transfer sensitive RF signals, power, other signals, pulse, whatever the client needs. The barrier must be certified to witstand a full defibrillator discharge that went wrong for some reason. That's 5kV out of a 32uF capacitor, thus potentially deadly.
One of the items used is a certain type of mil-spec aircraft wire to wind the transformers. There is an elaborate procedure that goes with that and it includes stock room requirements that I generally outline in great detail in my module specs to the client. All personnel in the stock room must be trained to touch this spool only when needed and should it ever be bumped, fall or whatever, this rather expensive spool must be discarded. Because the Teflon insulation could now be scratched, dinged, dirtied or otherwise compromised.
This means no personnel must be working this room that isn't trained. Hence nobody else is allowed access. Same goes for the production area while these transformers are manufactured. Nobody else past that yellow line.
When we designed this facility, we gutted it to the walls and started over. We decided that the stockroom would just be shelves in the production area, no walls. The only lockable doors inside the building are on the bathrooms.
No amount of checking can replace a process that doesn't allow the mistake to happen in the first place. Elementary quality control.
Who said fifteen years? It could be fifteen minutes.
Mistakes happen, particularly with uncontrolled documents get into the hands of engineers. ;-)
I don't much like you procedures (though I'm finding the ones at my CPoE are
*far* worse - I thought the Japanese were all over Demming, et.al.).
I hope he doesn't let you ECO it. You'll be twenty years in Club Fed. ;-)
No, you're ECOing an uncontrolled document. I'm not buying your *perfect* checking. I've never seen that.
Government? Yeah, that makes me sit up and listen. Rrriiigghtt.
board.
No, we're not medical (PTL).
Yes. You're going to check all netlists, how often?
Why? It is the document that the P-n-P machine uses.
Final test passes it, or not. The repair guys should know how the process works. Schematics laying in their desks are always a source of errors. Don't do that.
*Rarely* is the value of a resistor needed to understand the functioning of a circuit. If it is, chances are that it didn't change. QC uses a reference component placement drawing. It's notoriously bug ridden because it's another of those manual processes, but it is under ECO control anyway. :-(
It works and cleaned up a constant source of errors.
but, but, boss, no more respins because of board errors!
The repair guy is the only one with a set of schematics. He's smart enough to know how the ECO system works, at least enough to grab a BOM.
No, it *works*. We were doing it your way and constantly had errors creep in. Making a *clear* delineation between schematic=>board and BOM=>assembly cured all of one class of errors. It makes sense, when you consider the purpose of the schematic and BOM in the design process.
That's precisely the point. Those errors are now gone. Management likes that.
So having to slosh through BOMs to figure out which value goes into R92 is more safe? You've got to be kidding.
You are changing designs every 15 minutes? :-)
That's why there's the ECO process.
It's not really my procedure, it's the procedure used in many businesses that are under constant regulatory scrutiny because of the products they make.
It's a procedure that helps me not make mistakes. I bet if an IRS inspector would see that he'd be impressed. My CPA was. When he had a question I pulled out a binder and in around 10sec whe had the answer.
Not
Nope, we are not.
What can possibly go wrong in a netlist comparison? The netlist cannot change until you spin the board. Part values is where mistakes can happen and that is no different in your solution where the BOM can be wrong. In our cases these are usually validated by several people independently.
your
Whether you like it or not, in such markets they are the guys deciding whether you can sell product or not. My clients prefer to sell product.
board.
Then hope nothing bad ever happens.
Once, but with more than one person involved. We don't have any problems with that and do not find this task difficult at all.
absolutely
They are not looking at machines, they are looking at human document interfaces. At the board repair station, for example. Without exception every company I've ever been at has a repository of board schematics there from which they'd pull the one they need, or LAN access. You can bet money on it, that is among the first things an inspector will go through. To see that it's all properly released stuff without hand-edits.
Nowadays it all gradually migrates towards all electronic versions. Then they may audit the access mechanism, to make sure the repair techs can only access what is released.
Huh? We make sure they have current, released schematic that 100% match the units they are working on. So there are no errors on those. That's IMHO the only way to run a proper and efficient board repair station.
If you don't give them proper schematic, they might :-)
In my line of work they do need component values, a lot. "Why is there only 1.9V? It doesn't compute ... S..t! Someone changed R92 from 12.1k to 14.3k but the schematic says 12.1k!"
So whenever that happens you want them to slosh through the BOM and find out whether or not any of the resistors in the vicinity of this node have changed, and to which value? That is going to be a colossal waste of time for the techs.
Our QC used schematics. They had EEs who understood those.
The formal release of consecutive schematic versions can hardly cause a board error, can it?
the
And then? Painstakingly enter all changes into the old schematic? What a waste of his time. Plus, in our biz, that gets the place shut down by the Federales if they find out.
Then you did it wrong :-)
Being forced to carry the old original schematic version makes no sense to me. None at all.
the
As I said, then you have to hope your board repair techs never screw up so bad that something bad results. Those chances are much lower with the method used in me devices companies because they have current and released schematics available. More current information -> less mistakes.
Than having boards out of sync, you bet your a$$. As I said, it's a rarity that the tech would need the value to understand the circuit. He may measure the wrong value, looking for an easy-out, but he'd better check the BOM before he replaces parts!
Sometimes it seems like it. ;-) None lasted 15 years. I doubt many lasted one without an ECO.
...and that's why we ECOd your sucko ECO process. ;-)
With a lot of baggage from days when the schematic didn't directly create the board. I'm running into that where I am now. I just found out that they won't let me show shields/grounds and even thermal pads, on schematics! It has to be done by the layout guy! Stupid!
I bet there isn't a release schematic to match the process. ;-)
Not
It is. BTDT.
^^^^^^^^^^^^^^^^^^^^^^^^^^
You answered for me, above. ;-)
your
Whether *I* can sell a product? Nope. Not yet, though we do sell to Government Motors. ;-)
board.
Huh?
We found it impossible to control.
absolutely
The schematic is another source of error. The BOM is correct, by definition.
OK, but you'd better take all the pencils out of the lab.
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.