Analogous to the software world where every program has at least one bug and can be re-written in a better way with fewer lines. So therefore every software program can be reduced to one line of code that doesn't work!
Mark.
I agree with your philosophy John... but you do allow yourself a non-zero number of blue wires or a couple of tack-soldered components or something on those saleable rev. A boards too though, right?
That's great!
I do, but much less frequently than a decade ago. I recently thought about adding a hot air station but when thinking harder I realized that I would use it way too little to justify the desk space it'll take.
But he has tons of lab space. Heck, they've even got a crane in the lab. For lifting heavy equipment and transformers around all I've got is Ibuprofen, in case back pain creeps up :-)
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
Some people evidently get close enough. If you don't try to get it right on the first time you never will, though. The problem is when management expects to get it right on the first try, then demands it on the second. About the fifth...
What oven? We have a lot of problems with QFNs and interboard connectors (no problems with LEDs or inductors). We had one of the IE profs in from the local university to take a peek. He said we were probably pushing the capabilities of our reflow oven, particularly with RoHS crap. He also took samples of the parts to do surface analysis (our inventory techniques need a close look too). I think we're going to end up with a better oven. The owner doesn't want to spring for it, but if it's really the problem it's possible that we could double our production. He *would* like that. ;-)
It's really the same with hardware and software: Well-intentioned, conscientious [*], hard-working. competent people do make mistakes and you often can not catch your own because you see what *should* be there not what *is* there. Code/schematic/layout reviews, formal or informal, really are necessary. (I know I'm preaching to the choir here.)
[*] holy crap... I spelled that correctly!-- Rich Webb Norfolk, VA
We do design brainstorming, informal reviews (yey, wouldja look at this for me?) and a formal review after the board is laid out. It's surprising how many dumb mistakes get caught, and the occasional serious mistake.
Cose doesn't get reviewed much, because it can be changed any number of times without pads falling off boards. It has to be tested pretty well.
John
My way is to use a solderless breadboard, but build up sub-circuits on home-brew PCBs. So I've got a USB adapter, MCU adapter, ethernet socket with discretes adapter, power supply boards, etc. Homebrew can whip up a whole panel of breadboard adapters for SOT TSSOP CSP etc in little time and at little cost. The breadboard is for interconnecting the modules, adding pullup/pulldowns, etc. Once the circuits are explored this way, it goes to a PCB fab as a single final board (no production runs for me, just a hobby) and if it doesn't work I hack it until it does.
Here's an example of that process (scroll down to the breadboard pics):
We have even better, a technician. ;-)
SNIP
The current oven is made by TWS Automation in Italy and is pre RoHS. 2 guys can move it.
It was OK after moving to RoHS but I am packing bits tighter and using larger SMT parts such as 12mm square inductors which suck up the heat.
I had a guy going through test failures today. I saw a big pile with the same fault. I was told the oven was killing SOT-223 transistors. First I had heard of it. Some one will die on Monday.
When I found the LED problem all LEDs where replaced on existing boards, about 600. They are now put on by hand which is a PITA but I know they are reliable.
Now I know I have boards out there with suspect transistors:(
I had to get LEDs put onto aluminium backed boards elsewhere as there was no chance of this machine coping.
The new beast, 1.3 tonnes, is a second hand Ersa Hotflow-5 made in Germany and cost about twice as much for the new TWS one, £12k.
Spec here:
We had to upgrade ventillation and electrics for another £2k.
If CA was a bit less onerous with rules and regs and taxes I might as well. But ...
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
How do you get those nice clean cuts into the copper?
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
_THE_ professional world? Joerg, Joerg, you've been holed up in that mountain lair of yours for too long. ;)
Simulate the parts that simulators get right, do the rote stuff by rote, but prototype the stuff you're not sure will work. It's amazing the amount of stuff you can learn in a short time from a dead-bug prototype.
If you're just talking about laying out boards for circuit prototypes, then I agree--you might as well try a bit harder and get it right the first time. But trying out weird stuff, especially in mixed-technology systems, really needs prototypes.
Besides, lots of my protos are actually small instruments that I build in half a day and then use for years. An example is the sub-Poissonian current source and LNA I built for my tunnel junction work--very specific, worked great for years, took a day all told to design and build. Good medicine.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
made
how
to a
o you
d in
t to
.gamate
ction
ds
lose
You're getting a bunch of replies from naysayers who have gotten it "right" the first time, but AFAICT mostly from people who have done projects or one-offs.
In the world of production though, that's the exception rather than the rule IME. Even if the circuit does exactly what you first daydreamed it could do without a single glitch, even if whoever etches the board doesn't wire a pot backwards (I've had that happen), at some point in the product development cycle somebody will alter a spec just enough so that just enough redesign is required that your baby needs Frankensteining. It might be you didn't include enough LEDs for the required "ooh, shiney" level, the case design asshat^H^Hartists decided the air vents are in the wrong places, it has to go "boop" instead of "beep" when junior feeds it a PB&J, or whatever.
SOMEBODY will find a reason it needs "fixing".
Mark L. Fergerson
These days, I'll Spice it then go to schematic and layout. If I'm lucky, the board is ready for production. If not, I can usually test out most of my circuitry on the bum board.
If it's a tricky circuit, like a wierd switching power supply which requires good layout practices, I'll do a prototype circuit board and run it through one of the quicky circuit board places like
For digital projects, my colleague will do a test run on Xilinx place and route to catch I/O pin definition issues before committing to layout. For high-speed digital layout, treat it like an RF analog board.
If you understand Spice, it is a valuable tool. Bad part models can get you in trouble. Rohm's transistor models I used were/are minus the junction capacitances. I knew something was wrong when the response was flat out to 10MHz.
Sometimes I'll dead bug MSOP and similar packages, but only for small circuits. I need to be hard up to do this. You may consider surfboards to convert surface mount to through-hole. Digikey carries these things.
These days, I have a zoom stereo microscope ($500), hot air plate for warming the board to 150 deg C ($155 + $30 in modifications), and a hot air station for soldering ($185). All equipment is cheap Chinese stuff which works fine for repair and prototyping small parts. Our labs are equiped with Metcal soldering irons which is the place not to skimp. I don't use solder paste on leadless packages as it is too hard to dispense on small pads (e.g. 8-lead 1.6 x 1.6 mm packages). Easier to apply a bump of solder to the PCB pad and coat the underbelly of the part with flux, then hot air solder.
-- Mark
Most of our first boards go in to production without cuts and jumpers. These are usually moderately complex boards.
Now, now, we do have a modern feed store in this here town whar we're gitten them alfalfa bales and all that, and they even use a computation machine :-)
Ok, I did build a breadboard for my first noise-critical fiber-optics front end but that was more because the client really wanted that done. I ended up not changing a thing on there and going straight to a multi-channel layout. Since it has digital delay controls with SPI and stuff it (almost) counts as mixed signal.
One-off things I also often build on experimental board. I am not a great friend of the dead-bug style, preferring Vector board with a ground plane. That's harder to find these days so I stocked up. Many things go into those little Pomona boxes that end up riding on the back of a coax connector. All good medicine, but at least I can put a shiny aluminum lid on it so the clients don't see the wire ball inside my probes.
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
How do you get the flux back outta there?
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
I've recently gone from the solderless breadboard - protoboard - PCB cycle to breadboarding the bits I don't know and CADing everything else. Motivation: even for one-offs, it's easier to move components around in CAD space than desoldering and moving things in real life. Layout is a lot easier when you already know where everything goes, the error rate is lower, and it simply looks better.
Two boards so far, fairly large as hobby boards go (~500 holes), one worked perfectly first time (a switching power supply similar to your average computer PSU). The other had a few errors I missed on schematic entry, and a few other errors due to negligence. Nothing big, some flying wires and cut traces fixed it. 99% was fine, so it was a worthwhile prototype.
Tim
-- Deep Friar: a very philosophical monk. Website:
amalgamate
Very common situation when you are a consultant. New client, first article should have shipped a couple months ago but it still doesn't work. Everyone shaking in their boots. _Then_ your phone rings, agreement is signed same day, you call off any and all birthday parties and whatnot and you've got exactly one shot to get it right.
They'd have me over the barrel :-)
-- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
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.