tangent
Top Mall-Fi poster. The T in META42.
Formerly with Tangentsoft Parts Store
- Joined
- Sep 27, 2001
- Posts
- 5,969
- Likes
- 58
Someone recently emailed me a set of questions that are way too open-ended and general for private email. I prefer that such answers go to public places so that more than one person gets to learn the answers, and others can chime in with their own thoughts.
I'm putting this here because the person who asked came to me from Head-Fi as a result of another recent set of posts here. I'll let that person remain anonymous, since they did not intend for their questions to become public. I've anonymized their questions to some extent.
To get into it is easy: go download a board design package. There are at least half a dozen majors, and a bunch more small players.
Competence and mastery are entirely different questions.
Before you place your first PCB order, I’d say you’d better watch at least a few hours of YouTube/Vimeo tutorials, then expect to spend your free nights for a week designing it. And by that I mean something relatively simple, not one week to a flawless PIMETA v2 clone.
If you’re going the EAGLE path, my old tutorials are still relevant, even though they were done with EAGLE v4.
If you’re going the KiCad route, Chris Gammell’s tutorials are probably your best bet.
To ground this discussion, let's consider the PIMETA v2. It probably took me several months of nights and weekends to do the schematic and PCB layout work. I was working from a foundation laid over several years of similar work. The META42 was my first PCB design project, and it, too, took months of nights and weekends.
Several of these projects were done at least partly in public, and some were collaborations. These two forces pushed each project further than it would have gone if I was the only one driving it. I recommend that you emulate this mode of work, if you can. Go ahead and cook up the basic design in private, but once you've got something you think might work, post it publicly for comment. The iteration that results will get you to a better result, faster, than you could have accomplished solo. These comments will try to pull you in multiple different directions, which aren't all the way you would have gone on your own, but you will find that some of these directions are interesting and valuable. Choose a few to pursue, and incorporate them into the design. Repeat until done.
When you look at something finished like the PIMETA v2, you're seeing such experience distilled. What you aren't getting is the personal experience of having gotten to that point. Whether you know it yet or not, the PIMETA v2 is not the shallow end of the pool. Start with something simpler.
Cloning Gerbers is easy. Creating something more interesting than a clone of something that someone else created is work; lots of it.
How much? I’d say that depends on your starting skills and proclivities. Someone who gets easily confused by spatial transforms will be likely to make silly layout errors that result in a bad board. No one is a "natural" at PCB design, though different people will find it easier to achieve mastery than others.
This mastery is as difficult to acquire as in any other cognitively-demanding field that is complicated and valuable enough to be a full-time profession. That is to say, it takes about 10,000 hours of dedicated, skill-building practice.
The best single book I know of for getting into this particular topic is Peter Wilson’s The Circuit Designer’s Companion.
I have the 2nd edition by Tim Williams. I doubt there's anything that's truly unique in this book, something present nowhere else. What makes this book valuable is that no other book collects all of it into one single place. It is also uncommonly well-presented and designed. Within its scope, it is on par with AoE. It is not a replacement for AoE, but they make fine companions.
The Williams/Wilson book is pretty expensive for its size. Amazon wants about $72 for a copy today; I paid about $55 for my 2nd edition copy in 2005. The new edition adds about 125 pages. The delta also covers 10 years worth of consumer price inflation and paper book market collapse. $72 seems fair in that light.
That said, you might want to pick up a used copy of an older edition instead. This is evergreen material. They still make PCBs from glass-epoxy in 2015, just like they did in 2005.
It varies greatly. It depends on:
I’m not going to recommend board houses. I’ve been out of the game too long, and anyway, I never had a great loyalty to any of them. I chose the manufacturer for each board order based in large part on details about the particular order, which won’t be relevant to your situation.
There are too many to choose from for me to tell you what the best option is for you.
I've used four different packages over the years, and have played with more. And even then, I've never touched some really popular alternatives.
If you're on Windows, you pretty much have the pick of all options. Very few PCB design packages run only on non-Windows platforms. If you're on Linux or OS X, your set of choices will start out more constrained, which can be a good thing.
The hobbyist world has pretty much settled on EAGLE and KiCad now. I've spent most of my time with EAGLE, and have only started to use KiCad. I've put my early thoughts on KiCad vs EAGLE here.
Your own personal considerations will play a key part in choosing the best package for your needs. For example, maybe you're in some sort of school right now, where Altium has a big presence, so you can get a copy cheap. On the flip side, maybe you look down the road and see what an Altium license will cost you when you get out of school and decide instead that you really don't want to get hooked on such an expensive piece of software.
I'm putting this here because the person who asked came to me from Head-Fi as a result of another recent set of posts here. I'll let that person remain anonymous, since they did not intend for their questions to become public. I've anonymized their questions to some extent.
To get into it is easy: go download a board design package. There are at least half a dozen majors, and a bunch more small players.
Competence and mastery are entirely different questions.
Before you place your first PCB order, I’d say you’d better watch at least a few hours of YouTube/Vimeo tutorials, then expect to spend your free nights for a week designing it. And by that I mean something relatively simple, not one week to a flawless PIMETA v2 clone.
If you’re going the EAGLE path, my old tutorials are still relevant, even though they were done with EAGLE v4.
If you’re going the KiCad route, Chris Gammell’s tutorials are probably your best bet.
To ground this discussion, let's consider the PIMETA v2. It probably took me several months of nights and weekends to do the schematic and PCB layout work. I was working from a foundation laid over several years of similar work. The META42 was my first PCB design project, and it, too, took months of nights and weekends.
Several of these projects were done at least partly in public, and some were collaborations. These two forces pushed each project further than it would have gone if I was the only one driving it. I recommend that you emulate this mode of work, if you can. Go ahead and cook up the basic design in private, but once you've got something you think might work, post it publicly for comment. The iteration that results will get you to a better result, faster, than you could have accomplished solo. These comments will try to pull you in multiple different directions, which aren't all the way you would have gone on your own, but you will find that some of these directions are interesting and valuable. Choose a few to pursue, and incorporate them into the design. Repeat until done.
When you look at something finished like the PIMETA v2, you're seeing such experience distilled. What you aren't getting is the personal experience of having gotten to that point. Whether you know it yet or not, the PIMETA v2 is not the shallow end of the pool. Start with something simpler.
Cloning Gerbers is easy. Creating something more interesting than a clone of something that someone else created is work; lots of it.
How much? I’d say that depends on your starting skills and proclivities. Someone who gets easily confused by spatial transforms will be likely to make silly layout errors that result in a bad board. No one is a "natural" at PCB design, though different people will find it easier to achieve mastery than others.
This mastery is as difficult to acquire as in any other cognitively-demanding field that is complicated and valuable enough to be a full-time profession. That is to say, it takes about 10,000 hours of dedicated, skill-building practice.
The best single book I know of for getting into this particular topic is Peter Wilson’s The Circuit Designer’s Companion.
I have the 2nd edition by Tim Williams. I doubt there's anything that's truly unique in this book, something present nowhere else. What makes this book valuable is that no other book collects all of it into one single place. It is also uncommonly well-presented and designed. Within its scope, it is on par with AoE. It is not a replacement for AoE, but they make fine companions.
The Williams/Wilson book is pretty expensive for its size. Amazon wants about $72 for a copy today; I paid about $55 for my 2nd edition copy in 2005. The new edition adds about 125 pages. The delta also covers 10 years worth of consumer price inflation and paper book market collapse. $72 seems fair in that light.
That said, you might want to pick up a used copy of an older edition instead. This is evergreen material. They still make PCBs from glass-epoxy in 2015, just like they did in 2005.
It varies greatly. It depends on:
- Quantity.
There are usually steep price breaks at round numbers, on a logarithmic scale: 1 piece, 10, 25, 100, 250… You could pay $40 for a single board, but then $4 each for 250 pieces of the exact same design, from the same board house. Why? Because there’s a huge amount of overhead that goes into the initial production, and once that’s out of the way, it’s kind of like hitting “print” again to get more.
- Country of origin.
Boards made in China will be cheaper than boards made in Silicon Valley. It's a combination of local workforce costs, local real estate costs, local material and utility costs, etc.
Some of the most interesting board houses are in the light industrial outskirts of some big city — e.g. Denver, Chicago — where leases are more reasonable than in Silly Valley, but where there is enough local business that they can keep costs low. Most of my boards were made in such a place.
- Volume.
Boards made in a facility that caters to your order's sort of volume will be cheaper than in a board house that normally makes fewer or more boards per batch than you want. There are board houses that cater to the onesy-twosy market, board houses that are set up specifically to produce 3-10 piece prototype orders, board houses that specialize in short-run production quantities (100-250), and high-volume board houses (500+).
Sometimes it makes sense to pay more to get your prototypes from a board house on the higher-volume end of the scale so that you can avoid the doubled tooling costs that result from moving a design from one board house to another. That is, if you know you will want 100 pieces eventually, getting your prototypes from the place you intend to place that order might be worth paying a bit more for those first few.
- Turn time.
You can pay through the nose to get a board back the next day, or you can choose to sit on your hands for a month if you want your boards cheap.
- Finish options.
The cheapest sorts of boards may not have any solder mask, or if they do, it will be a single color picked by the manufacturer. There may or may not be silkscreen, you might only be able to get it on one side of the board, and uncommon colors typically cost extra, if they're available at all. The exposed pads may be solder-plated, silver-plated, gold-plated, or bare copper; each affects the cost.
A board with a black solder mask, double-sided yellow legend, and ENIG pad finish will cost a lot more than a board with a green mask, white top silk, and HASL pads, all else being equal.
- Quality.
The cheapest boards are often worth exactly what you pay for them, and no more. Really cheap boards may not be able to meet your requirements for trace spacing, trace size, solder mask accuracy, etc. Cheap boards often simply look ugly.
There are a few places that do cheap, high-quality, low-volume orders. Usually you give up on choice or turn time to get this. That is, you typically can't change the finish options and it may take weeks to get the boards back.
- Testing.
Expect to add $100-200 to an order to get each board electrically tested. This can make sense for dense designs ordered in quantity, so that you’re adding only $1 or less to each board’s cost, and saving yourself the pain of stuffing a bad board.
I’m not going to recommend board houses. I’ve been out of the game too long, and anyway, I never had a great loyalty to any of them. I chose the manufacturer for each board order based in large part on details about the particular order, which won’t be relevant to your situation.
There are too many to choose from for me to tell you what the best option is for you.
I've used four different packages over the years, and have played with more. And even then, I've never touched some really popular alternatives.
If you're on Windows, you pretty much have the pick of all options. Very few PCB design packages run only on non-Windows platforms. If you're on Linux or OS X, your set of choices will start out more constrained, which can be a good thing.
The hobbyist world has pretty much settled on EAGLE and KiCad now. I've spent most of my time with EAGLE, and have only started to use KiCad. I've put my early thoughts on KiCad vs EAGLE here.
Your own personal considerations will play a key part in choosing the best package for your needs. For example, maybe you're in some sort of school right now, where Altium has a big presence, so you can get a copy cheap. On the flip side, maybe you look down the road and see what an Altium license will cost you when you get out of school and decide instead that you really don't want to get hooked on such an expensive piece of software.
