Concept of laying out devices onto a PCB?

Thread Starter

careless_monkey

Joined Jan 11, 2011
17
I just couldn't find a straight forward answer to this question while searching google.
How does one make all the layout for components on the PCB? How does one know if a resistor goes at point A and then a capacitor or inductor at point B? And then having millions and trillions of these + transistors/op-amps and other devices in the circuit.
I do have knowledge of these devices but making them work together in combination of millions or trillions is something I just cannot grasp.
So can anyone explain me how?
 

thatoneguy

Joined Feb 19, 2009
6,359
Generally, signals move in a "direction", such as input to output, which lays out the rough guildelines.

From there, when hundreds of nets (connections between components) exist, there is software that auto-places and auto-routes the traces, sets up the ground planes, adds vias where needed, etc.

It can also be done by hand, which is a dying/lost art for advanced layouts, but still done often for basic layouts.

With high speed digital, there's always at least a ground plane, a power plane, and 2 (or more) signal planes. Essentially, several etched boards laminated together. Each layer has a different "preferred orientation", e.g. top layer the traces are up-down, bottom signals, the traces are routed left-right. This reduces capacitive coupling. For inner layers, 45 degree left and 45 degree right "direction preference" are given to tracks.

Signal paths are kept as short as possible, except in a bus, where they are both as short as possible AND the same length, so the signal arrives at the same time.

These rules worked backwards help decide where components go, as well as what orientation they should be in, such as which way is pin 1. passive components are frequently used as jumpers across data lines, which is the origin of the infamous "zero ohm resistor". There are entire courses and seminars on PCB Design and production, as this aspect involves a lot of RF rules to prevent signals from mixing. Designing a PCB for a high speed digital circuit is often as challenging as designing the actual circuit to start with.
 

Wendy

Joined Mar 24, 2008
23,415
Generally when I'm doing it I use the schematic as a reference. I put the ends of the device close as practical to the device it will be connecting to. This isn't a hard science, and you can adjust positions later (I do). Usually when I move things around after I start laying traces it is to get myself out of a bind.
 

retched

Joined Dec 5, 2009
5,207
Wow, you both started your posts with "Generally" Thats kinda weird.. ;)

Generally, I try to use set pre-set spacing and layout rules and let the software "keep me sane"

When things get "thick", at least I know that I am not out of spec for traces and pad size
 

thatoneguy

Joined Feb 19, 2009
6,359
Wow, you both started your posts with "Generally" Thats kinda weird.. ;)

Generally, I try to use set pre-set spacing and layout rules and let the software "keep me sane"

When things get "thick", at least I know that I am not out of spec for traces and pad size

It's close to impossible not to generalize when speaking of PCB layout. Unless you are one of those savants than can mirror everything in your head as a 3d model and simply draw what you saw.
 

marshallf3

Joined Jul 26, 2010
2,358
A good program will help out a lot for a simple circuit ( I personally use http://www.diptrace.com/ ) but when it comes to creating the actual PCB design I prefer to manually place my components in the most logical layout with the majority of the inputs on the left and outputs to the right. The power supply will find its own place as board space permits providing you've included plenty of bypass caps. You can use the program for free (200 or 500 pins for an EDU or non-profit status) or try out the full version for free.

Remember that your design may or may not need to be be mirrored, ages ago I made a board that came out backwards and while it was fine it was not what I wanted.
 
Last edited:

tom66

Joined May 9, 2009
2,595
I route all PCBs by hand, in a PCB program.

Autorouters suck in my honest opinion, especially when you're trying to minimise size or cost.



That is Super OSD Lite, one of my projects.

My tips are to modularise everything: I designed the OSD side, then the power supply, then the microcontroller, then the buttons, and had space to include an EEPROM and flash memory.

Routed entirely by hand in "PCB".
 

mcgyvr

Joined Oct 15, 2009
5,394
There is a lot of engineering judgement that goes into a proper professional circuit layout. Its not just where the components fit but things like thermal/rf interference considerations,etc... And even manufacturing issues that must be looked at. Wrong component placement can cause LOTS of problems. Things like the design guides/application notes for components usually have good information on layout considerations,etc... Of course this is more of a hobbiest site with hobby projects so it's not nearly that big of an issue but... I can't tell you the pains that arise when you need to get a product through professional EMC/EMI/FCC testing.


Talking about auto-routing.. I've found the newest version of Diptrace to do an excellent job of auto-routing (well for my boards anyways). There is always some manual work that must be done but I always start with the autoroute and just touch up any issue I see manually. Of course you must set it up properly too.. Garbage in..garbage out.
 

tracecom

Joined Apr 16, 2010
3,944
Talking about auto-routing.. I've found the newest version of Diptrace to do an excellent job of auto-routing (well for my boards anyways). There is always some manual work that must be done but I always start with the autoroute and just touch up any issue I see manually. Of course you must set it up properly too.. Garbage in..garbage out.
As a total novice to DipTrace and while laying out a relatively simple board, I found that the first step that worked best for me was to manually route crucial traces. I then allowed DipTrace to route nets one at a time while constraining the routing to one layer of the PCB. During this process, I made some manual corrections to some of the nets as they were auto-routed (such as rotating components.) Then, as a last step, I looked for (and found) several opportunities to reduce overall trace length by moving selected trace segments to the opposite board layer. As a result, I avoided vias since all the layer changes took place at component through holes.
 

tom66

Joined May 9, 2009
2,595
There is a lot of engineering judgement that goes into a proper professional circuit layout. Its not just where the components fit but things like thermal/rf interference considerations,etc... And even manufacturing issues that must be looked at. Wrong component placement can cause LOTS of problems. Things like the design guides/application notes for components usually have good information on layout considerations,etc... Of course this is more of a hobbiest site with hobby projects so it's not nearly that big of an issue but... I can't tell you the pains that arise when you need to get a product through professional EMC/EMI/FCC testing.
Agreed. Especially when dealing with power supplies, I follow manufacturer's application notes to the finest detail. With high-speed logic, I keep traces as short as possible, where that's feasible. And I group most signals together, so that there are no or minimal time differences between them.

Talking about auto-routing.. I've found the newest version of Diptrace to do an excellent job of auto-routing (well for my boards anyways). There is always some manual work that must be done but I always start with the autoroute and just touch up any issue I see manually. Of course you must set it up properly too.. Garbage in..garbage out.
PCB has an autorouter. It works reasonably well... but a human is better. For bigger layouts, it's almost inevitable that you'll have to use an autorouter, but smaller layouts like Super OSD only take me 2-3 hours to route.
 

Thread Starter

careless_monkey

Joined Jan 11, 2011
17
Thanks for all these comments!
I would like to learn more about laying out PCBs and building boards that actually do something. What would be the best way to do this? I'm doing ECE and it's my last year. But until now it's all been theory and nothing practical.
I was thinking of building a robot from scratch but having no practical experience in building boards, instruments, this project seems too much out of reach.
Can anyone suggest anything for me that can help me stamp a solid foot on the ground in this area?
 

tom66

Joined May 9, 2009
2,595
Thanks for all these comments!
I would like to learn more about laying out PCBs and building boards that actually do something. What would be the best way to do this? I'm doing ECE and it's my last year. But until now it's all been theory and nothing practical.
I was thinking of building a robot from scratch but having no practical experience in building boards, instruments, this project seems too much out of reach.
Can anyone suggest anything for me that can help me stamp a solid foot on the ground in this area?
Wow, that sucks... How can you just do electronics theory? When it comes down to it, electronics rarely follow theory precisely. And you can't work on theory alone. Unpredictable things not modelled can and do happen.

First, I'd recommend listing exactly what you need your project to do - figure out how big the robot will be, for example. The mechanical part will be tricky - figure that out and design it first. Then, start considering what you need the electronics to do, and create the schematics for them. Finally, after all that, you can begin the PCB layout - but it takes time to reach this point, in my case, it took 3 months, and I went through about 5 designs before settling on this one.
 

Thread Starter

careless_monkey

Joined Jan 11, 2011
17
Wow, that sucks... How can you just do electronics theory? When it comes down to it, electronics rarely follow theory precisely. And you can't work on theory alone. Unpredictable things not modelled can and do happen.

First, I'd recommend listing exactly what you need your project to do - figure out how big the robot will be, for example. The mechanical part will be tricky - figure that out and design it first. Then, start considering what you need the electronics to do, and create the schematics for them. Finally, after all that, you can begin the PCB layout - but it takes time to reach this point, in my case, it took 3 months, and I went through about 5 designs before settling on this one.
There were labs in the courses but that didn't help much to be true.

Firstly, I'll just be getting it to move around on the ground and in different directions. I still am looking on what is the easiest way in doing this. I was looking at DC motors and controlling that using a potentiometer. But, for it to actually move around in different directions, one has to actually sit on the device and change the pot. This is not possible and I moved onto servos and controlling them using radio like the ones used in RC or Toy planes. I'm still learning about them.
I have about 4 months from now to complete it and it is just an extra curricular project that I'm doing along with the courses I'm taking in my last semester
 

retched

Joined Dec 5, 2009
5,207
Can anyone suggest anything for me that can help me stamp a solid foot on the ground in this area?
The best suggestion is: Dont expect to have a "solid foot on the ground" any time soon.

As with any "art", there are seemingly unlimited ways to do the same thing.

There are trace extending methods that are used so all 8,16,32,64,etc. bits arrive to the next logic point at the same time.

There are on-board inductors, made using traces, there are ground planes, hatched and solid,

blah blah blah.

It all "matters" in one way or the other. If you REALLY want to become an "artist" at PCB layout, start looking at EVERY SINGLE PCB you can get your hands on.

Also, download a PCB software and practice it to no end.

You can grab a schematic off the net, input it into the PCB program, then start routing.

You will understand why things are the way they are by doing it yourself.

Also, keep in mind, that most everyone goes for shortest fattest tracks possible.

With that in mind, get started.

You will learn most "tips of the trade" by looking at professionally done PCB's.

It is an art. If you started painting TODAY, you would not expect to be Van Gogh tonight.

Google PCB layout tutorial

read, read, read
fail, fail, fail

and learn from it all.

Try to enjoy it, and have fun.
 

kubeek

Joined Sep 20, 2005
5,794
When I start placing and routing a PCB, I usually start flipping the parts around and placing them quite far away, then I run the autorouter to see if it is even possible to route it and to find problematic areas. After that, some more flipping, shrinking and autorouting until it looks quite good, then I usually route it manually.
 

tom66

Joined May 9, 2009
2,595
There were labs in the courses but that didn't help much to be true.

Firstly, I'll just be getting it to move around on the ground and in different directions. I still am looking on what is the easiest way in doing this. I was looking at DC motors and controlling that using a potentiometer. But, for it to actually move around in different directions, one has to actually sit on the device and change the pot. This is not possible and I moved onto servos and controlling them using radio like the ones used in RC or Toy planes. I'm still learning about them.
I have about 4 months from now to complete it and it is just an extra curricular project that I'm doing along with the courses I'm taking in my last semester
You won't want to use a potentiometer, because they dissipate a lot of heat, and are big. You'll want to use some kind of motor controller. It would simplify the design if you could stick with only one or two motors.
 

Thread Starter

careless_monkey

Joined Jan 11, 2011
17
When I start placing and routing a PCB, I usually start flipping the parts around and placing them quite far away, then I run the autorouter to see if it is even possible to route it and to find problematic areas. After that, some more flipping, shrinking and autorouting until it looks quite good, then I usually route it manually.
What software do you use to do this?

You won't want to use a potentiometer, because they dissipate a lot of heat, and are big. You'll want to use some kind of motor controller. It would simplify the design if you could stick with only one or two motors.
Yes I have two motors at the moment. I will be connecting them by a shaft. They'll also be used to steer.
 
Top