PCBs correct first time?

[Leon Heller]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Leon

 

[Roger Hamlett]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Pretty true.
However the reasons nowadays, are not usually through lack of checking the
design. Typical problems would be unexpected noise (even the best simulators
still can't model things exactly like the real world), or a component that
displays tolerances in production quantities that were not displayed on the
original samples.
A couple of recent boards I did (reasonable high density, four layer double
eurocard size), had one problem on the first, and two on the second. One
problem in each case was from the same chip (which was a thermal problem
when built in the final casing), and the other was was an induced noise
problem from a motor mounted adjacent to the board. Finding these, is what
prototypes are all about.
What has basically 'vanished', in the last few years, are the
'transcription' errors between circuits and the final board, and 'silly'
faults like tracks crossing one another. Rats nets checkers, and basic
design rules, fix these.

Best Wishes

 

[raymund hofmann]

I think you meant "Designs correct first time" - As a PCB usually is
checked against the netlist of the schematic, and mechanical
design-rules.
So if it isn't correct it may be a "design error" or some "mechanical
error" or some "physics" (RF, etc) design error not covered by netlist
verification. But this also could be described with and checked against
design rules.

From the projects I do I would say that nearly no prototype is correct
the first time. And even if there is a prototype doing all things
required by the customer the first time I will take some time to play
around with it for "optimization". This helps keeping the "Models" in my
Brain in sync with the prototypes.
And often the confrontation with the Prototype brings some new
ideas/requirements to the customers mind.

But it also depends on how "new" each Project is. If it's like building
a car already built, but painted in a new color, then chances for a
first time correct design are very high.

But if your "social succes" is measured by your "first time success
rate" I would suggest learning strategies like.

- justify errors so that they are no errors anymore.
- avoiding doing something new, but concentrating at suggesting others
that it is new
- rhetorically blame others for your blatant errors.

Raymund Hofmann

 

[nospam]

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

If you can't get it right first time when you try, just how bad is it going
to be when you don't bother trying?

 

[Frank Bemelman]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Only a *very* small PCB has a chance of being 100% correct
or at least useable. There are always little things that
go wrong. New footprints that are wrong, little mistakes
in the schematic.

I check with automated design rule checking, that doesn't
cost much.

Even the second proto often has a few minor things that
can be improved. Like the silkscreen layout, for instance.

 

[Eugene Kaplounovski]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Leon

What do you mean by 'one error'? If this is the layout not matching
the schematic, than the person simply doesn't know how to use their
CAD package - all modern CAD can check layout against the netlist
better than any human, and the post-processing can easily be verified
by numerous viewers (like GerbTool in OrCAD for example). Of course, a
responsible designer should never trust somebody else's footprints
(including the ones bundled with the CAD package, they are often the
worst).
If, on the other hand, it is the schematic that needs correction,
that's (partly) what prototyping is for.
Regards,
Eugene

 

[budgie]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Many replies posted here rely on DRC and netlists to ensure correct
translation from schematic to PCB layout. As one who uses Protel
Schematic and AutoTrax under DOS (remember DOS?), I rely on the Mk1
eyeball for all checking - along with checklist procedures I developed
yeras back.

The last commercial design I did went straight from PCB file to
production. There was ONE error, due to wrong pinout information for
a concentric DC input receptacle. Fortunately, it only required a
bridge between two adjacent lands to fix. The client was so
unconcerned that he ran several more batches through production before
even bothering to have the artwork revised. Needless to say, the
Rev:1 board had 0 errors.

IMOE the source of many problems is wrong pinouts or footprints in
libraries, or in published data sheets.

 

[EEng]

How do you breadboard a design with 50 or so fine-pitch parts, a uP, a
couple of 650-pin BGAs running at 80 MHz, and a lot of analog stuff?

Even for fairly simple stuff, you'd probably spend more time and money
breadboarding than going directly to a PCB. The real danger in
breadboarding is that you, well, design by breadboarding.

John

There are many options today that didn't exist even 10 years ago. SMT
protoboards are great for testing some designs, and breadboarding
isn't expected to be actual finished size. As far as BGAs and other
parts of like ilk, most companies have adapters to allow them to be
breadboarded, you have but to ask. Remember that breadboarding
doesn't prove PCB layout, it only proves schematic design. After
that, PCB layout is only as good as the person doing the layout, and
anyone with experience knows about EMI, thermal dynamics, etc.....
Good layout technique cannot really be taught; it is learned by
experience.

Anyone that designs by breadboarding is not worthy of the position.
Breadboarding is for proof of concept, NOT as a design method.

 

[Tom Del Rosso]

In budgie typed:

Many replies posted here rely on DRC and netlists to ensure correct
translation from schematic to PCB layout. As one who uses Protel
Schematic and AutoTrax under DOS (remember DOS?), I rely on the Mk1
eyeball for all checking - along with checklist procedures I developed
yeras back.

Spell checkers do have their uses though. ;-)

 

[budgie]

In budgie typed:

Spell checkers do have their uses though. ;-)

Fortunately, that was the result of sloppy typing, not dyslexia ;-)

 

[Leon Heller]

I think Leon meant get a prototype board made, not your whole
production run. At least, I hope he did!

Yes, that's what I meant.

Leon

 

[Ian Buckner]

I read in sci.electronics.design that Boris Mohar

The problem is often, IME, that the footprint is for a ecived nip 41
(-
--
Regards, John Woodgate, OOO - Own Opinions Only. http://www.jmwa.demon.co.uk
Interested in professional sound reinforcement and distribution? Then go to
http://www.isce.org.uk
PLEASE do NOT copy news posts to me by E-MAIL!

Good call! - one that nearly got us was an assumption about which
way was "up" between a motherboard power connector and the
connector on our board at the other end of a wiring harness, with
the two engineers sitting 6k miles apart.

Regards
Ian

 

[The Real Andy]

No one in my experience (especially me) ever seems to get PCBs of any
complexity designed correctly straight off, there's always at least one
error on the prototype. They are usually OK the second time round, of
course. How general is this observation?

I've got a feeling that thorough checking of the prototype PCB design
probably costs more ultimately than just getting it made and fixing any
problems subsequently.

Leon

My PCB's are always correct the first time. Its always the environment that
is wrong...Oh, a don't forget those pesky little quirks that the IC
designers forget to mention in the data sheets

 

[Frank Bemelman]

My PCB's are always correct the first time.

In your dreams

Its always the environment that
is wrong...Oh, a don't forget those pesky little quirks that the IC
designers forget to mention in the data sheets

It would be nice if that's the only problem.

 

[Boris Mohar]

In your dreams


It would be nice if that's the only problem.

Mounting holes? What mounting holes?
Test points? What test points?

 

[Spehro Pefhany]

In your dreams


It would be nice if that's the only problem.

I have yet to (knock on wood) have even a prototype come back that was
not shippable with a bit of rework. I've even ordered 1,000+ fairly
expensive boards without a prototype. This may change as the cost of
prototypes has dropped so much that it may be cheaper to spend less
time checking, but I don't really think so. The issues like mechanical
dimensions, 3-D fit, and so on are easier to check with laser prints
and spray adhesive to laminate test bits. Footprints on parts that I
don't have samples in hand are a nasty issue, but I try to always have
samples around to lay on top of the laser print (crude, but effective
check). Then there are possible schematic issues, but that's nothing
to do with the layout. It's sometimes good to highlight the power nets
on the screen, both to check they are going to all the places you
expect, and to make the way the current will flow easier to visualize.

I expect the prototype to be about as good as the first production
units used to be, and then the actual production parts ought to be
close to perfect. Sometimes I don't bother fixing stuff like silk
screen markings over vias until production, but that risks
accidentally getting them over SMT or other pads (but I run two
independent DRC checks to catch most of that stuff). Of course if you
made the footprint wrong, the padstacks wrong, or selected the wrong
one...

Best regards,
Spehro Pefhany

 

[Frank Bemelman]

I have yet to (knock on wood) have even a prototype come back that was
not shippable with a bit of rework. I've even ordered 1,000+ fairly
expensive boards without a prototype. This may change as the cost of
prototypes has dropped so much that it may be cheaper to spend less
time checking, but I don't really think so. The issues like mechanical
dimensions, 3-D fit, and so on are easier to check with laser prints
and spray adhesive to laminate test bits. Footprints on parts that I
don't have samples in hand are a nasty issue, but I try to always have
samples around to lay on top of the laser print (crude, but effective
check). Then there are possible schematic issues, but that's nothing
to do with the layout. It's sometimes good to highlight the power nets
on the screen, both to check they are going to all the places you
expect, and to make the way the current will flow easier to visualize.

I expect the prototype to be about as good as the first production
units used to be, and then the actual production parts ought to be
close to perfect. Sometimes I don't bother fixing stuff like silk
screen markings over vias until production, but that risks
accidentally getting them over SMT or other pads (but I run two
independent DRC checks to catch most of that stuff). Of course if you
made the footprint wrong, the padstacks wrong, or selected the wrong
one...

But you are a brilliant guy, Spehro! I always **** up something. My
latest screw-up was not having a footprint for a PLCC44 PTH socket.
Couldn't find one with google. I had a PLCC32 PTH socket, and modified
it, turning it into a PLCC44 socket. Of course, with pin 1&2 starting
in the wrong position. At the moment of creating that thing, I was
sure I got it right. I could use the board for all other prototype
testing, with a socket dangling of the board, with 44 wires. You
don't want 1000 boards of that

The trouble is, it all looks so bloody nice on the screen. Zero
complaints from the DRC. And you upload the files, happy as a
monkey with 7 dicks.

 

[Spehro Pefhany]

But you are a brilliant guy, Spehro! I always **** up something.

No I just put lots of time into checking and fixing what gets
forgotten. It works better if you can set it aside and look at it
later, or get someone else to check it. It's almost always something
really stupid and obvious, like using the wrong width for a SOIC. But
it's probably only a matter of time (knock on wood).

My
latest screw-up was not having a footprint for a PLCC44 PTH socket.
Couldn't find one with google. I had a PLCC32 PTH socket, and modified
it, turning it into a PLCC44 socket. Of course, with pin 1&2 starting
in the wrong position. At the moment of creating that thing, I was
sure I got it right.

I'd probably go back and check that about 4 times. Those thru-hole
PLCC sockets have a really irritating pin pattern (I remember making
an 84 or 68 footprint). Then there are relays and oddball connectors
with metric dimensions, asymmetrical layout, pinout numbering
including various phantom pins, and the drawing given from the
*bottom*. Yuk.

I could use the board for all other prototype
testing, with a socket dangling of the board, with 44 wires. You
don't want 1000 boards of that

That's when you get the boards right out in the trash so you don't
have to be reminded of the monumental screw up.

The trouble is, it all looks so bloody nice on the screen. Zero
complaints from the DRC. And you upload the files, happy as a
monkey with 7 dicks.

Hahah. I haven't heard that expression in years. Then there was the
guy with only five- but his underwear fit like a glove.

Best regards,
Spehro Pefhany

 

[John Devereux]

I expect the prototype to be about as good as the first production
units used to be, and then the actual production parts ought to be
close to perfect. Sometimes I don't bother fixing stuff like silk
screen markings over vias until production, but that risks
accidentally getting them over SMT or other pads (but I run two
independent DRC checks to catch most of that stuff). Of course if you
made the footprint wrong, the padstacks wrong, or selected the wrong
one...

I just did a board with only a couple of minor errors:

o A 176 pin QFP with 0.5mm pin pitch. Except it should have been
0.4mm.

o A flash chip with all the address lines off-by-one.

All in all, not my best work...

 

[Frank Bemelman]

I just did a board with only a couple of minor errors:

o A 176 pin QFP with 0.5mm pin pitch. Except it should have been
0.4mm.

o A flash chip with all the address lines off-by-one.

All in all, not my best work...

Hahaha, I once used the PLCC schematic symbol for a dualported
ram, 7C136, and used the PQFP footprint. Pin1 starts on the corner,
not in the middle. I ditched it right away, about 5 minutes after
that first and exciting 'power-on' event.

You're not alone