Zero to Expert in pcb design

[Curious.George]

Re: Zero to Expert in pcb design

I used to do the whole project myself from concept on napkin, to mechanical design based on Industrial design concept that the marketing people want, schematic capture, PCB layout, create part library as needed, part sourcing, BOM, wiring and assembly instructions (I build the unit myself first to see how difficult it is will be for the production people to build the products and make adjustment as needed based on the feedback from production people), Compliance testing, test procedures from board level test to finished product test, user manual validation, documentation for ISO, etc. until the company is bought out by big company, so now I am just a number in the big corp but I am going to try to retire by the end of this year after 25 years with the same company.

I live within the needs/resources of the client. They obviously aren't keen on paying me (outrageous wages) to do something that THEY can do. So, I defer to whatever their budget (and capabilities) can accommodate. E.g., if they've got an assembly house that can build prototypes, then they'll do that instead of paying my wages to "be a Tech".

[More than once, I've had to beg off on clients who want me to "train" some of their staff: "Sorry, I'm not in that business. Buy suitable people!"]

The biggest "disconnect" is usually in the tools that I have available vs. what they are willing to invest in. E.g., I don't think twice about dropping $10K-50K on an asset that I can leverage for a future job; they, OTOH, might see this as an extra expense that they may never be able to use, again.

So, it gets tricky -- if I use "expensive tools" to do what they need done, then that tends to limit them to using me for any follow-up work, or, someone with similar, compatible tools (note that I tend NOT to want to do followup work as there's very little to be learned/gained in those exercises). OTOH, if I don't use those tools, they typically have to pay for more hours to get the same job done -- and they STILL may be dependent on "someone" if they don't have the right skillset(s) in house. (and, I may not be interested in their job if I can't use it to polish my skills on some tool -- I've no desire to start cutting Rubylith just because that's where their process is!!)

Enjoy your retirement! I live in fear of the day I can't keep doing what I'm doing, presently (there's something about the pressure of having to get a project done that makes it different from projects that you just want to tinker with, without that pressure).

[budm]

Re: Zero to Expert in pcb design

I used to do the whole project myself from concept on napkin, to mechanical design based on Industrial design concept that the marketing people want, schematic capture, PCB layout, create part library as needed, part sourcing, BOM, wiring and assembly instructions (I build the unit myself first to see how difficult it is will be for the production people to build the products and make adjustment as needed based on the feedback from production people), Compliance testing, test procedures from board level test to finished product test, user manual validation, documentation for ISO, etc. until the company is bought out by big company, so now I am just a number in the big corp but I am going to try to retire by the end of this year after 25 years with the same company.

[Curious.George]

Re: Zero to Expert in pcb design

I work closely with mechanical guys, for the from factor requirement during layout to make sure all will fit together and will also pass the compliance (UL, TuV, FCC, etc.), and Design for manufacturability which is the big part of the whole thing, plus cost.

I tend to work with small firms with tiny/nonexistent Engineering departments. This is a win, for them, in that they're only paying me when they genuinely NEED "engineering" services. It's a win, for me, in that I don't have to take on jobs/tasks that aren't of interest to me, personally (or my business goals).

A downside of this is that there is rarely any prior art/tools to inherit; I have to build everything from scratch -- hence my focus on having tools that make creating parts easy (symbols and footprints).

An upside of this is that I can ensure every part I make (regardless of client) shares a common approach -- and, that approach has some rational basis! I am so tired of symbols that resemble the physical device pinouts instead of presenting the signals in a functional/logical manner (i.e., signals presented in numerical order by pin number instead of grouped by function, regardless of pin number!)

We use Altium in conjunction with Solidworks on all 7 engineering locations across the globe, we have big library that we build up for so many years.

I've systematically purchased a wide variety of tools to ensure I can address every aspect of product development in a (reasonably) effective manner. I don't chase the latest releases (EDA/CAD tools are pricey!) but, rather, live with older releases unless I have a pressing need to upgrade (or, a serendipitous opportunity presents!)

As there's more to product development than schematic capture and PCB layout, I have tools that let me address each of those other areas, as well. E.g., I write a User Manual before designing the product; and, a Service Manual as soon as the paper design is complete (i.e., before there is any real hardware to photograph for the manual's illustrations). This sort of approach lets stake holders review the product before it's actually been fabricated to see if I've missed some feature or design aspect that's important to them.

[I'm presently learning about designing hard tooling for injection-molded parts and how that impacts the cosmetic aspects of the device. A "big company" would no doubt have an IE tackle that job and I'd be waiting on his "vision" before I could make significant progress]

[budm]

Re: Zero to Expert in pcb design

I work closely with mechanical guys, for the from factor requirement during layout to make sure all will fit together and will also pass the compliance (UL, TuV, FCC, etc.), and Design for manufacturability which is the big part of the whole thing, plus cost. We use Altium in conjunction with Solidworks on all 7 engineering locations across the globe, we have big library that we build up for so many years.

[Curious.George]

Re: Zero to Expert in pcb design

BTW, beside learning how to use CAD, you also need to learn about how the traces should and should not be routed. If you look at some of the application notes of the of the devices, they will also give you the board layout guide lines, it is not just connecting point A to point B. Most PCB layout person will work closely with circuit designers and with more senior guys that have lots of real world experience.

It's not just "electrical" layout issues (matching trace lengths, controlled impedance runs, trace widths for specific current handling requirements, minimum track spcing, etc.). You also have to address placement issues (if certain components NEED to be in certain locations), potential mechanical interference (e.g., if there's a rib in the plastic case into which the board fits and that rib imposes a keep-out zone on the PCB in that area), thermal issues (power transistor leaning up against a 'lytic), air flow (a wall of tall components can block air from passing), manufacturability (clearance for the PnP), serviceability (where's a good place for a ground tie? and which TPs should be added?), etc.

Altium, Flotherm, etc. tackle some of these issues for the layout guy. But, only if you invest the time in setting up the job to use those capabilities. If you're just doing one-offs or a little "garage shop" business, its not likely you'll go to those extremes.

[Sadly, many small businesses just fly by the seats of their pants. I've encountered several who've lost the schematics for the designs they are producing (or, have schematics that are out of date), lost the source code for MCU-based designs, have no formal processes for tracking changes to released product, etc. They fail to realize that their biggest assets are those documents, not the pile of parts in the stockroom!]

Good layout guys know most of this sort of stuff and just need "special instructions" for things that they can't glean from an examination of the design (cuz most aren't EEs).

If you do then I do not understand why you ask question in post #3.

Hell, I'd like a tool that did that for me!

[budm]

Re: Zero to Expert in pcb design

BTW, beside learning how to use CAD, you also need to learn about how the traces should and should not be routed. If you look at some of the application notes of the of the devices, they will also give you the board layout guide lines, it is not just connecting point A to point B. Most PCB layout person will work closely with circuit designers and with more senior guys that have lots of real world experience.
BTW, so already have all these software: Altium, Mentor PADS, Proteus and Orcad and have been using them already?
If you do then I do not understand why you ask question in post #3.

[Curious.George]

Re: Zero to Expert in pcb design

Wow, ouch, I am way too old to become an expert like you guys, I won't live that long.

Rule of thumb: 10,000 hours to become "expert" on something. So, roughly 5 or 6 years of focusing 100% of your time on it.

So , back to the subject here, the learning curve on most of the software I got now, Altium, Mentor PADS, Proteus and Orcad is fairly simple, but thanks to this thread I got a small taste of what knowledge the experts are dealing with.

If you can grasp the intricacies of PADS or Altium (beyond etch-a-sketch mode), you're pretty much "there". Each imposes their own idea of how they should be used (instead of letting YOU decide how you want to use them). That's the biggest hurdle, IMO, to this sort of software -- some "programmer" having a notion of how he wants to WRITE the software ends up dictating how YOU will end up USING the software! :<

[It's also one of the reasons why it is so tedious to move a design from one vendor's tool to another -- even by the same vendor!]

Ultimately I want to see if there is an area that I can break through in electronics software design and revolutionize it to do things the experts only dream of.

Tools + Application Domain Experience = Results

Typically, you end up with one or the other and miss the mark. Like electronic products that are impossible to service.

[captainKKK]

Re: Zero to Expert in pcb design

Wow, ouch, I am way too old to become an expert like you guys, I won't live that long. I am just a geek who desires to learn enough to be dangerous and accomplish what I dream up. So , back to the subject here, the learning curve on most of the software I got now, Altium, Mentor PADS, Proteus and Orcad is fairly simple, but thanks to this thread I got a small taste of what knowledge the experts are dealing with. Ultimately I want to see if there is an area that I can break through in electronics software design and revolutionize it to do things the experts only dream of. You can think of it like this.... many revolutionary software improvements in the design of CAT crackers in the petroleum industry were made because only someone that understands the process and hurdles of chemical engineering, and also have the skills in software will be able to make these changes. Thanks.

[Curious.George]

Re: Zero to Expert in pcb design

i have fine probes and patience,
every pin/leg gets buzzed to every other pin/leg on the board.
yes, it takes a while, but your not effected by hidden via's or boards with intentionally obfuscated tracks hidden under a flood-fill groundplane on each side.

I have nowhere near that amount of patience -- nor TIME! (And, clients would cringe at my hourly rate for such a menial task)

Here's a pic of the first board I "scanned", many years ago (the design is now almost 30 years old). A crude count shows (about):

• 3 DIP40
• 4 DIP28
• 5 DIP24
• 13 DIP20
• 18 DIP18
• 28 DIP16
• 22 DIP14
• 6 SIP10
• 240 2-terminal discretes
• 20 3-terminal discretes
• (neglect connectors)

So, about 2,200 nodes in ~150 sq. in. of board -- that's not even 1 DIP16 per sq in... incredibly low density, by any standards! (my last thruhole design was 2 DIP16's per sq in; I now do SMT work, exclusively) Laughably so compared to SMT designs!

2,200 items, taken two at a time (combinations, not permutations) would mean probing 2,418,900 sets of points, recording the result and transcribing to real components on a schematic.

At 1 per second, that would take 672 hours (remember, you actually have to USE the data you are collecting from these probes).

Granted, you can make semi-intelligent guesses as to which pins you NEED to probe for any given starting point, but that leaves you vulnerable to overlooking some potential connection (and, besides, YOU claimed to patiently probe every pair). And, each such assessment requires time to consider.

If it was something as banal as a PC motherboard, you could probably make huge assumptions about "what's likely to go where" -- cuz they are largely similar to each other. OTOH, the boards in your TV are very disimilar from each other and neither resembles any of the boards buried in your vehicle or microwave oven.

Having refined my tools and skills (in addition to technological advances in the COTS tools I use), I suspect I could have a fresh new schematic from scans of this board in less than two days -- ready to prepare a new layout on the third day (I'd need to consult a stuffing list to fill in the values of all of the discretes; the actives are legended in the silkscreen). I imagine it would take a draftsman longer than that just to capture the schematic from "paper".

There are firms that now offer this as a service -- though I have no idea what their rates are (likely high as they are competing with ad hoc and "highly manual" methods)

[stj]

Re: Zero to Expert in pcb design

i have fine probes and patience,
every pin/leg gets buzzed to every other pin/leg on the board.
yes, it takes a while, but your not effected by hidden via's or boards with intentionally obfuscated tracks hidden under a flood-fill groundplane on each side.

[Curious.George]

Re: Zero to Expert in pcb design

it's been reliable for me manually probing and building a netlist - i'v reversed some pretty large boards with many dozens of chips on them.

So, how do you find EVERY pin to which a particular net connects -- "hope"?

You must have great digital dexterity and eyesight to hold probes on SMT pads without them slipping off and touching a neighboring pad (which would give you "bad data").

I trust the machine to get those things right so I don't have to waste my time (and risk inaccuracies in the result).

[stj]

Re: Zero to Expert in pcb design

it's been reliable for me manually probing and building a netlist - i'v reversed some pretty large boards with many dozens of chips on them.

[Curious.George]

Re: Zero to Expert in pcb design

you dont trace foils, you create a netlist from the component pins.

Without resorting to building a custom, one-off test fixture of pogo pins to electrically probe the (depopulated) PCB, can you tell me another method of doing that -- other than what I've described?? :>

Or, do you think manually probing every combination of pin-pairs with for continuity is a RELIABLE way of doing this?

[stj]

Re: Zero to Expert in pcb design

you dont trace foils, you create a netlist from the component pins.

[Curious.George]

Re: Zero to Expert in pcb design

Humm, Diptrace Full is only $995....I wonder if its Schemat Capture allows users to take a picture of any board and it will produce a likely pcb with components....? Is there anything out there with this capability? I like OrCad's licencing to VM, so I can take it anywhere.

I've hacked COTS packages to "assist" in building a schematic from scans of top and bottom layers of (4 layer) boards (the assumption being that internal planes only route power and that can be added to a schematic/netlist "by hand"). Basically, you ask the tools (schematic capture + layout) to identify "layout errors" based on the schematic that you've been incrementally building.

(This is where having LOTS of display real estate pays off! Open schematic on one monitor, layout on another and have a third/fourth to call up datasheets for specific components on the board)

This is 100 times better than tracing foils by eye. But, still increibly labor intensive! What it really buys you is some reassurance that you haven't confused yourself (again!) when you flipped the board over to see where a signal at a particular feedthru continues along...

[captainKKK]

Re: Zero to Expert in pcb design

You mean smart enough yet.....
How long do you think a chinese expert would take to conceive a new computer motherboard design and take it from design to production?
Did you know, Intel has a free tool called Embedded Board Planner? Through a web interface, you can for example design your own Intel Atom based processor board....https://www.fedevel.com/welldoneblog...6-motherboard/

[mariushm]

Re: Zero to Expert in pcb design

Nobody does that, because nothing's smart enough to automatically figure out the number of layers a PCB has, and x-ray capability to figure out traces within internal layers.

Also, big components like transformers or inductors or even electrolytic capacitors can hide traces under their footprints making traces undetectable.

[stj]

Re: Zero to Expert in pcb design

that's not what schematic capture means.

[captainKKK]

Re: Zero to Expert in pcb design

Humm, Diptrace Full is only $995....I wonder if its Schemat Capture allows users to take a picture of any board and it will produce a likely pcb with components....? Is there anything out there with this capability? I like OrCad's licencing to VM, so I can take it anywhere.

[Curious.George]

Re: Zero to Expert in pcb design

I kinda like Mentor Graphics PADS almost as well as Altrium. Proteus also has a lot to offer.

Colleagues using Diptrace have had good things to say about it. YMMV

You have to decide what you want from the package. If you just want to layout some boards (and don't care about back annotating the schematic, etc.), then you have lots more choices than if you're trying to put together a whole package for Manufacturing.

My designs tend to be more complex and dense (6-8 layer padstack, components on both sides of the board, schematic sliced into several tiny boards which are "folder" together, etc.). So, I need the tools to handle lots of details that I could easily overlook.

For example, when the second "pane" of the board is folded behind the first, will any components on the BACK of the first pane mechanically interfere with components on the back of the second pane?