Circuit Board Design: Basics of Designing a PCB

[The_Inventor]

My background is in physics so I have a very basic understanding of circuit design. I was just wondering, If someone were to design a printed circuit board (from scratch) how do they know that it will function the way it should? and also how do they know that the printed circuit board will satisfy all the conditions for it's use? For example if I wanted to build a calculator or a digital watch from scratch I would need to use software to make a schematic of the PCB but, how do I figure out how to arrange the components so that the PCB actually does what I want it to and better yet how do I figure out what components to use? Sorry if this was long and confusing.

 

[davenn]

I was just wondering, If someone were to design a printed circuit board (from scratch) how do they know that it will function the way it should?

in it's very basic form, the circuit schematic can be directly translated to a PCB layout by making sure there are tracks that connect the same components to each other as shown in the schematic

OK a basic circuit

a basic PCB for that schematic

you can clearly see where each component goes when comparing the circuit to the PCB layoutDave

 

[davenn]

and also how do they know that the printed circuit board will satisfy all the conditions for it's use?

what conditions specifically ? temperature and water resistance ??
what other things were you thinking of ?

For example if I wanted to build a calculator or a digital watch from scratch I would need to use software to make a schematic of the PCB but, how do I figure out how to arrange the components so that the PCB actually does what I want it to

for the more complex layouts, many of the PCB design programs will take a schematic and will create a board for that schematic
with little or no tidying up needed

The PCB drawing programs will also allow a user to place components and await your instructions on what should be connect to what else

better yet how do I figure out what components to use?

well that happens a long time before worrying about a PCB
Learning how to design circuits is a lifetime of experience ... I have been at it for many decades and I still don't know
all the fine details, there are those on this forum that are far above me when it comes to design
A good understanding of electronic theory and how components work is required ... study ... study ... study !

I will freely admit, I am not much of a designer beyond the basics
but give me the schematic and the components and I will build it ( up to a certain level of complexity)Dave

 

[Rive]

Sorry if this was long and confusing.

I have to admit, it really is confusing. Are you really asking us that what is 'engineering'?

For circuit design it's not really different from any other 'engineering'. You have to know what do you want: what do you have: and what will they do when you put them together. Then figure out a solution.
The PCB is just an advanced way to connect stuff.

 

[phinds]

For example if I wanted to build a calculator or a digital watch from scratch I would need to use software to make a schematic of the PCB ...

No, you would not HAVE to. When I started out as an EE circuit boards were laid out by technicians working from a circuit diagram.

 

[Averagesupernova]

My assumption is that the original poster here is looking at a finished circuit board and making some very broad (and incorrect) assumptions. Almost everything should be known about the end product long before the circuit board is designed and built. It has to make sense to the designers on paper in schematic form first. This has to mesh with any software running on a microcontroller that may be part of the design.

 

[Stuschreib]

Circuit design, and especially PCB design, is a sort of "black art", like Voodoo. Textbook circuit design methods are used to choose "what components to use", but experience will tell you which circuit "topology" will strongly depend on precise component values and parameters, and which are more tolerant to component variation. When it comes to PCB design, simple, "slow" circuits are almost trivial, but as signal speed increases, as in high speed digital and radio frequency circuits, all sorts of effects start to be significant. While first principles will give guidance on how to deal with these effects, again it is experience that will overcome them.
It is not uncommon for a complex, high speed circuit to NOT work properly on the first try, even if designed by a pro. Computer simulation helps, a lot, but in the end, a hair-tearing, trial-and-error approach will be needed. That's the difference between a ME and an EE; the ME actually expects his/her product to work as designed on the first build.

 

[berkeman]

If someone were to design a printed circuit board (from scratch) how do they know that it will function the way it should? and also how do they know that the printed circuit board will satisfy all the conditions for it's use?

One of the features of your schematic drawing software package is the ability to output a "Netlist" that shows in a standard format what all the components are (their sizes and their "pad pattern" for the PCB layout), and how the pins of the components are connected to each other. The PCB Layout software package Imports this Netlist, and will create a "Rat's Nest" layout with the parts just dropped onto the PCB outline, with the connections shown. Then you rearrange the components to make the best layout you can, to minimize trace cross-overs and logically put parts together more like they are shown on the schematic.

There are lots of variations on this basic flow, but hopefully that helps you get the idea. There are also simulation software packages (mostly based on the core SPICE package) that let you simulate the operation of your circuit before you actually build it. That let's you optimize component values and make sure there are no fundamental errors in your circuit design.

Here is a sample Netlist from a recent small design of mine at work:

% LPT-11 SIP                                    Revised: Thursday, September 08, 2005
% 012-0317-02                                   Revision: A
%
%
%
%
%
BOARD = ORCAD.PCB;

PARTS
SMT0805         = C1,   % 1.0uF
                  C10,   % 0.1uF
                  C11;   % 0.1uF
SMT0603         = C12,   % 0.1uF
                  C13;   % 0.1uF
SMTTANTA        = C14;   % 4.7uF
SMT0603         = C15;   % 0.1uF
SMT0805         = C16,   % 1.0uF
                  C17;   % 0.68uF
SMT0603         = C18,   % 47pF
                  C19;   % 47pF
SMT0805         = C2,   % 0.68uF
                  C20;   % 0.1uF
SMT0603         = C3,   % 0.0033uF
                  C4;   % 0.1uF
SMT0805         = C5,   % 0.1uF
                  C6,   % 1.0uF
                  C7;   % 0.1uF
SMT0603         = C8;   % 0.1uF
SMT0805         = C9;   % 0.1uF
SMT_SOT-416     = D1;   % DAN222
SOT-416         = D2,   % DAN222
                  D3;   % DAP222
SOT-323         = D5;   % DAN202U
SOT-23_RR       = D6,   % CMPD2005S
                  D7;   % CMPD2005S
LPT11EDGECON_14 = P1;   % EDGE_CONN
SOT-323         = Q2,   % BC847W
                  Q6,   % BC847W
                  Q8,   % UMT2222
                  Q9;   % BC857BW
BC847B_SOT-363  = QA1;   % BC847BDW
SOT-363         = QA10,   % BC847BPN
                  QA3;   % BC847BPN
BC847B_SOT-363  = QA5;   % BC847BDW
SMT0603         = R1,   % 51.1k
                  R10,   % 4.53k
                  R11,   % 100.0k
                  R12,   % 61.9k
                  R13,   % 71.5k
                  R14,   % 3.92k
                  R15,   % 10.0k
                  R16,   % 51.1k
                  R17,   % 100.0k
                  R18,   % 7.15k
                  R19,   % 5.11
                  R2,   % 100.0k
                  R20,   % 324k
                  R21,   % 619
                  R22,   % 324k
                  R23,   % 2.0k
                  R25,   % 100.0k
                  R26,   % 5.11
                  R27,   % 5.11
                  R3,   % 7.15k
                  R4,   % 5.11
                  R5;   % 1.0k
SMT0805         = R6;   % 3.48 Ohms
SMT0603         = R7,   % 18.2k
                  R8,   % 1.50k
                  R9;   % 4.53k
TP35            = TP1,   % TEST PT
                  TP10,   % TEST PT
                  TP11,   % TEST PT
                  TP12,   % TEST PT
                  TP13,   % TEST PT
                  TP14,   % TEST PT
                  TP15,   % TEST PT
                  TP16,   % TEST PT
                  TP17,   % TEST PT
                  TP18,   % TEST PT
                  TP19,   % TEST PT
                  TP2,   % TEST PT
                  TP20,   % TEST PT
                  TP21,   % TEST PT
                  TP22,   % TEST PT
                  TP23,   % TEST PT
                  TP24,   % TEST PT
                  TP25,   % TEST PT
                  TP26,   % TEST PT
                  TP27,   % TEST PT
                  TP28,   % TEST PT
                  TP29,   % TEST PT
                  TP3,   % TEST PT
                  TP4,   % TEST PT
                  TP5,   % TEST PT
                  TP6,   % TEST PT
                  TP7,   % TEST PT
                  TP8,   % TEST PT
                  TP9;   % TEST PT
SMTSO8          = U1,   % TL5001A
                  U2;   % FTIC2

NETS

T1  = R26/2 U2/1 TP28/1 ;
TBIAS  = R15/1 TP20/1 U2/3 ;
T2  = R27/2 U2/8 TP29/1 ;
T2R  = D3/1 C11/1 TP18/1 C10/1
            D2/2 R27/1 ;
T1R  = TP12/1 D2/1 C9/1 R26/1
            D3/2 C7/1 ;
VCC  = C13/1 C3/1 C4/1 R7/1
            U2/2 D2/3 C6/1 P1/5 ;
N2  = TP27/1 D1/1 C6/2 R25/1 ;
NPB2  = TP2/1 QA1/4 C2/1 Q2/1
            R3/1 ;
NPB1  = C1/1 R2/1 QA1/6 R1/2
            QA1/5 TP1/1 ;
NPB3  = Q2/2 QA1/2 R4/1 TP3/1 ;
NNB1  = QA5/6 R17/1 R16/2 TP21/1
            C16/1 QA5/5 ;
NNB2  = R18/1 TP22/1 C17/1 Q6/1
            QA5/4 ;
NNB3  = R19/1 Q6/2 TP23/1 QA5/2 ;
GND  = R5/2 R10/2 R25/2 D5/1
            C13/2 R16/1 C12/1 R13/2
            P1/6 R12/2 C14/2 Q6/3
            C4/2 R15/2 U2/7 U1/8
            QA5/3 D5/2 QA3/3 C19/1
            D3/3 C5/2 R11/2 C15/2 ;
SCB  = C20/2 Q9/2 TP4/1 R6/2
            QA10/5 R23/1 Q8/3 ;
SCP  = R5/1 QA10/3 TP15/1 U1/5
            C15/1 ;
VO  = R14/2 U1/1 TP11/1 ;
B3  = R22/2 TP8/1 R13/1 QA3/5 ;
FB  = TP24/1 R10/1 U1/4 D1/3
            R9/2 R8/2 R7/2 ;
COMP  = U1/3 C8/1 TP16/1 ;
DTC  = C14/1 TP17/1 R11/1 U1/6 ;
RT  = TP25/1 R12/1 U1/7 ;
A2  = TP14/1 C8/2 R9/1 ;
Y  = TP26/1 Q9/3 Q8/1 R21/1 ;
B2  = R20/2 QA3/4 TP7/1 QA3/2 ;
VHV  = TP10/1 QA3/1 U1/2 QA10/2
            C12/2 ;
B4  = R23/2 QA10/6 TP6/1 Q9/1 ;
E11  = TP9/1 QA10/1 R14/1 ;
NET_A  = P1/1 C9/2 C7/2 D6/3 ;
N1  = TP13/1 C3/2 R8/1 ;
NET_B  = C11/2 D7/3 C10/2 P1/2 ;
V+  = QA10/4 C18/2 QA1/1 C2/2
            C5/1 R22/1 R4/2 R3/2
            C1/2 R2/2 P1/3 C20/1
            QA3/6 R20/1 R6/1 ;
CLK  = P1/7 U2/6 ;
TXD  = U2/4 P1/9 ;
RXD  = U2/5 P1/10 ;
INDUCTOR  = P1/4 Q8/2 R21/2 D5/3 ;
VBRIDGE+  = D6/2 TP5/1 QA1/3 C18/1
            D7/2 Q2/3 R1/1 ;
VBRIDGE-  = C17/2 C19/2 C16/2 QA5/1
            R17/2 R19/2 TP19/1 R18/2
            D7/1 D6/1 ;

 

[Baluncore]

If something is going to be manufactured then there first needs to be a working prototype or model. A designer does not have to start from scratch because the prototype will usually be an evolution of an earlier version. That goes back through many generations to someone's creative imagination. Many products contain redundant hardware because inept designers have progressively added new features by adding complexity, without understanding the reasoning behind the design. It is actually a mistake to remove something if it cannot be explained fully why it was placed there originally, as there may still be a critical situation that you have not yet understood. And there is never enough good design documentation.

These days, given a circuit schematic, there are computer programs that will do an OK job of laying them out. But it takes a most unusual person to do a good job of laying out circuit boards by hand and eye, even with computer drafting. They need to have an excellent 3D sense, 'feel' the pattern of signal flow and 'know' the sensitivity of different parts of the circuit to changes in PCB parameters. If they cannot read maps then they really don't stand a chance laying out PCBs. To some extent it is still an art, different designers have different styles. PCB design is a full time job with a steep learning curve. After several years designing circuits and PCBs, an engineer will have disappeared into some hidden cave, from where they will be watching component technology as it advances, then applying the useful ideas and parts to future projects and updates.

The aim of every design engineer must be to get a safe and functional product into the market place as quickly as possible. They then watch for changes in technology that might lead to potential reductions in cost of manufacture, improvements in reliability, or added features. At some point determined by economics, those changes will be prototyped and a new version or model of the product will come into existence. It is a continuous process. Whenever the engineer is not finalising a design for manufacture, they are mentally preparing to design future products.

 

[Billing Tom]

Thanks for sharing this article here. I need your help please. Hope you might know the source. I need PCB circuit board manufacturer. Where is this available. Actually I am looking for a reliable source.

 

[jim hardy]

I need PCB circuit board manufacturer. Where is this available. Actually I am looking for a reliable source.

I've used PCBExpress for simple prototypes. You design the board using their software and they mail it to you.

There are many many outfits in the PCB manufacturing business and a simple search will offer you a smorgasbord to choose from. Probably there's one near you.

old jim

 

[phyzguy]

My background is in physics so I have a very basic understanding of circuit design. I was just wondering, If someone were to design a printed circuit board (from scratch) how do they know that it will function the way it should? and also how do they know that the printed circuit board will satisfy all the conditions for it's use? For example if I wanted to build a calculator or a digital watch from scratch I would need to use software to make a schematic of the PCB but, how do I figure out how to arrange the components so that the PCB actually does what I want it to and better yet how do I figure out what components to use? Sorry if this was long and confusing.

Most designers will build a model of the circuit using SPICE before actually building it. I strongly recommend doing this, since you can build up the circuit and try out different components before you actually invest the money to build the circuit. LTSPICE is available for free download. You can draw the schematic using available components, then simulate the behavior. Once you are happy with it, you can go to a PCB manufacturer like @jim hardy mentioned and get the PCB built. For a small board these can actually be quite inexpensive. I've been told you can buy a small simple PCB for under $10.