Top Tips for Designing PCB's so they are Manufacturable.

[mrpackethead]

Pickit is slow on larger parts but has the huge benefit of programmer-to-go mode for standalone operation, which AFAIK isn't yet implemented on PK4 ( last time I looked there wasn't even any mention in the docs). 

yeah, its very slow.  Too slow to be useful, for anything other than if you are doing some in the field upgrade thing. I have used it to send a to a customer, in an attempt to make it super easy.    Though most times, customers are a long way away, and they perfer an update via file!

[palletguy]

Hi All, NEWBY here, with a different perspective on PCB design;

1st off, understand the electronics assembly industry and how a blank PC end's up being an assembled PCB used in a device, understanding how all the parts go together is essential for a good board layout & design. Think through the entire process...Then design the board.

Secondly: Regardless if your building the PCB yourself or Outsourcing, absorb every response that is given in this forum, yes including mine. Understand that proper spacing of all bottom & topside SMD to PTH is critical in the manufacturing of the card and the final price! Will the card need a solder pallet, not just for the SMT process but wave solder (including Select soldering & Batch soldering), AOI & X-RAY, Press fit, Conformal Coating, etc, etc... Everything you design today will have a major impact on the future build of this PCB and the people that build it, along with the costs associated with building this PCB. Remember this: HANDLING KILLS A CIRCUIT BOARD.

3rd: See & Feel the design: Look at the parts being placed & inserted on the card, what role will panelization play when building the card, what extra expenses may be required to build the card (pallets), Look downstream man!! Look past the billfold and create a design a circuit board that can be built in the least amount of time, with the fewest defects possible. It can be done!

When I started in the electronics assembly Industry many moons ago, 85% of the people I dealt with were circuit board designers. At that time (mid to late 90's) design engineers realized that if they were to create a circuit board using guidelines established by the solder tooling world, they could create a board that can start at one end of the plant, pass through all QC checks and placed in a box at the end of the day. That never happens anymore, once a year I might hear from a designer looking for some advice on the proper spacing between SMD & PTH, but that's it.

Understand the process of circuit board assembly. Once the design is complete, get the valued opinion of at least three people (Manufacturing & Process Engineers that will build the product) and see what they have to say!

Then design a circuit board.

[rx8pilot]

Hi All, NEWBY here, with a different perspective on PCB design;

1st off, understand the electronics assembly industry and how a blank PC end's up being an assembled PCB used in a device, understanding how all the parts go together is essential for a good board layout & design. Think through the entire process...Then design the board.

If that were the case....very few people would ever make a PCB.
I recommend....fail fast and fail often in the case of PCB design. In general, it is faster and cheaper to learn what you can, make a design, send it to the fab/assembly. Your mistakes will be called out and you will learn fast. The cost of the bad boards is less than trying to become an expert before you even try.

There are rules and regs that govern good PCB layout, but experience and intuition cannot be studied. Pick some easy-ish projects and make them. Get feedback from others and use the responses to make the next one better. Assemble them yourself as much as possible, there are a lot of lessons to be learned while staring at the PCB for 2 hours while you assemble it.

[cgroen]

Hi All, NEWBY here, with a different perspective on PCB design;

1st off, understand the electronics assembly industry and how a blank PC end's up being an assembled PCB used in a device, understanding how all the parts go together is essential for a good board layout & design. Think through the entire process...Then design the board.

If that were the case....very few people would ever make a PCB.
I recommend....fail fast and fail often in the case of PCB design. In general, it is faster and cheaper to learn what you can, make a design, send it to the fab/assembly. Your mistakes will be called out and you will learn fast. The cost of the bad boards is less than trying to become an expert before you even try.

There are rules and regs that govern good PCB layout, but experience and intuition cannot be studied. Pick some easy-ish projects and make them. Get feedback from others and use the responses to make the next one better. Assemble them yourself as much as possible, there are a lot of lessons to be learned while staring at the PCB for 2 hours while you assemble it.

Exactly!

[tggzzz]

Whats your top tips for designing PCB's that are manufacturable??   or improve cost and reduce faults.

There are many good points about PCBs containing modern logic in: http://cache.freescale.com/files/training/doc/ftf/2014/FTF-SDS-F0198.pdf

[djacobow]

If your prototype board doesn't work as it should, make it work by any means necessary (cut traces, bodge wires, dead bug fixes, etc) before you spin again. That is, do not be an idiot and just fix the first problem you found before you re-spin.

[mrpackethead]

Its certainly a lot easier than it used to be!

[NorthGuy]

I can't think of many scenarios where HV programming on MCLR is a problem

You have just received your new expensive PIC32 board, you connect it, you open the PIC32 project. However, the other PIC16F628A project is still open. You click the "make" button, the PIC16 project compiles, PICkit3 applies 13V to MCLR - bye bye PIC32.

The PGM pin is only needed for very old PICs - the new ones do not have it and still can do LVP.

[Bassman59]

If your prototype board doesn't work as it should, make it work by any means necessary (cut traces, bodge wires, dead bug fixes, etc) before you spin again. That is, do not be an idiot and just fix the first problem you found before you re-spin.

You mean "don't use Agile methods for hardware design."

[djacobow]

If your prototype board doesn't work as it should, make it work by any means necessary (cut traces, bodge wires, dead bug fixes, etc) before you spin again. That is, do not be an idiot and just fix the first problem you found before you re-spin.

You mean "don't use Agile methods for hardware design."

Yes. I've just seen it done so many times, particularly by non-HW people who have been tasked with hardware dev. They are perpetually surprised when there is more than one thing wrong with their boards.

On the other hand, this isn't really a DFM-specific issue. It's just a don't waste time and money issue, which is pretty basic engineering.

[T3sl4co1l]

I've been tasked with cleaning up a "REV G" board more than once...

Tim

[JPlocher]

The ends of a spectrum....
one end: hack & go @ $10/ea per iteration.  The other:  study, learn the theory, apprentice to the masters, grok the entire process

One gets you practical knowledge, a feeling for good enough, and a pile of drink coasters; the other, potentially, a product that can be produced economically at scale.

I'd guess that most of us are in the first category, trying to learn from the 2nd.

In many ways, good Design for Manufacturing is as complex as "feeds & speeds" and "tool & work holding" are in the shop - the concepts and practices are complex and hard for the novice to comprehend, until, that is, they have walked the path of enlightenment (aka coaster production) and see for themselves what that advice really means.

The competing wisdom in this thread would make a great "Zen of circuit board design" guide :-)

  -John

[mrpackethead]

The ends of a spectrum....
one end: hack & go @ $10/ea per iteration.  The other:  study, learn the theory, apprentice to the masters, grok the entire process

Unfortuantly its not $10/interation is it.  Its $1000's of dollars and weeks of time.

[T3sl4co1l]

I pride myself on doing "Rev B" designs: the first prototype is pretty close but missed a few things; the final build fixes exactly those problems and is ready for production.

Gotchas are sometimes my/our fault, but it's also often the case that the customer changes their spec -- it happens -- and then it goes up from there.  (We've got one customer that's pushing rev 10 on some boards, but they're also a long time customer and their product is slowly but constantly evolving.  That's fine, beats spinning a whole new product line after all. )

Tim

[mrpackethead]

My plans normally work on three interations;

(a) A prototype
(b) A revision
(c) Production.

Sometimes i go a->c.. and sometimes theres a b.1

[ar__systems]

Depends on the customer... Not always there is a luxury of having full specs from the start. On recent project the final revision is rev 12, of which I actually manufactured 4...    The last one was to fix EMI issues.

[mrpackethead]

Depends on the customer... Not always there is a luxury of having full specs from the start. On recent project the final revision is rev 12, of which I actually manufactured 4...    The last one was to fix EMI issues.

If the customer is paying then changes = increased costs to them. 

[AgileE]

If you ever get the chance to visit an assemblers factory you will get further insight into what can make boards easier or harder to put together. The grimacing from the operators is a fine indication of a problem board 

[mrpackethead]

If you ever get the chance to visit an assemblers factory you will get further insight into what can make boards easier or harder to put together. The grimacing from the operators is a fine indication of a problem board 

Bringing assembly into our own environment made us think very hard about our designs.  Seemingly minor changes can make a big improvment to how things work.