What is Circuit Design

Site: Teach Me Printed Circuit Board Design
Course: Learn Board Layout by Designing a Badge
Book: What is Circuit Design
Printed by: Guest user
Date: Tuesday, 9 August 2022, 4:44 PM

Description

Circuit Designers try to make connections.

1. What do PCB Designers Do?

Printed circuit board design is, at a very basic level, an activity that makes electrical connections between parts, pads, and pins.  But making connections is the easy part of the job.  If all that needed to be done was to connect pads and pins, there wouldn't really be a need to do anything more than create a schematic and let a computer program do the rest.

Bob Pease Book on Solderless Breadboard

The cover of Bob Pease's book "Troubleshooting Analog Circuits"

Designers have to worry about how components behave in actual circuits. Do they get too hot?  Are there electromagnetic field interactions between two signals?  How do you balance cost and quality?  How do you reduce the footprint by 3mm²?  How do you route 40 A without the board melting?  Will my 120VAC circuit arc and destroy something?

Printed Circuit Design is a decidedly creative activity that requires engineers to understand applied physics and problem solve.  There are competing priorities that must be constantly balanced, and one change can create a cascade of other changes.  For now, an experienced engineer is superior to a computer, but that might not always be the case.

While this course doesn't dig into the deeper layers of Signal Integrity and Power Integrity, the topics will be mentioned from time to time.  The following pages provide a bit of background on the circuit design process.

2. Design for Manufacturability

The Problem

Early electrical engineers designed for connectivity.  One of the original ways that the pins of vacuum tubes, wires, capacitors, and other components were connected was with wire-wrapping.  This technique requires a machine or operator to wrap a small wire around a post that protrudes through a non-conductive base.  Each net is laboriously constructed in a spaghetti-like mixture of wires and pins.

Wire wrap boards

Example of wire-wrapping from Wikimedia.org

As you can imagine, mistakes were common and almost impossible to trace.  Assembly, troubleshooting, and repairs were difficult if not impossible -- in some cases, it was faster to simply remove all the wires and start over.  That is simply not a sustainable production model.  As early as 1903, engineers began creating the precursors to modern circuitboards and by the middle of the 20th century, the circuit boards we recognize today began to appear in consumer appliances.

The Current Solution

The way the industry collectively decided to solve the problem of connecting the pins of the various microcontrollers is with interconnects made of horizontal copper traces to make in-plane connections, and electrodeposited copper to make out-of-plane connections.

Picture of a two sided PCB

This two-layer circuitboard is the power supply unit from a big-screen television.

This artistic image shows a combination of horizontal and vertical interconnects with the dielectric layers removed

It would be a mistake to assume that PCB manufacture will continue down this one path to manufacture.  There are companies working to integrate additive manufacturing techniques into the PCB space.  For now, this is the world we live in and one that electrical engineers must understand.

So, let's take a quick look at what it takes to put a modern PCB together.

3. Process Overview

The Basic Process

When you order a PCB from a fabrication house, an engineer specifically trained in PCB fabrication lays out all of the necessary steps needed to make the PCB and sends the steps to the production floor.  As the project moves from station to station and step to step, the complete design slowly takes shape.  Some PCBs can be made in a dozen steps, and others require hundreds of steps.  Each additional step in the process increases the material cost or the time required to make the PCB,  and every step increases the risk of failure and the overall cost of the PCB.

Example Traveller

This is the mockup of the first page of a multi-page traveler.

All PCBs are a bit different -- but there are a few processes that are common every board.  Depending on the board requirements, these steps can be performed in a different order or repeated for certain features.

3.1. Material Selection

Whenever possible, shops use prefabricated panels of copper-clad laminate material they purchase from suppliers. 

Warehouses at Royal Circuit Solutions

Shelves at Royal Circuit Solutions in Hollister, California are stocked with various copper-clad cores

These panels are typically 12"x18", 18"x24", or 24"x36" in size. Entire panels must be fabricated as a single unit - so to save costs, engineers will often order multiple projects or multiple copies of a single project on the same panel.

Panel in Electroplating Tank

This panel is coming out of an electroplating tank -- it shows 12 copies of a board on a single panel.

One way to save money in a design is to ask your fabricator if you can split your panel with another customer. In fact, entire companies such as OshPark have been built on this model.

Multiple Projects on a Single Panel from OshPark

This image shows multiple customer designs on a single panel.  The photograph is from https://twitter.com/oshpark

Copper Weight

The foil that is used on a PCB comes in several standard copper weights with 0.25 oz, 0.5 oz, and 1 oz used most frequently in standard PCB designs.  Heavy copper refers to 1.5 oz or greater.  Most shops keep up to 3 oz copper weight on hand.  Royal Circuits keeps 4 oz and some 6 oz on hand.  Heavier copper weights must be ordered from a supplier, which adds at least one day to a PCB order.

Electroplating increases the copper thickness of every board.  So a board made with 0.5 oz copper that is electroplated with an additional 0.5 oz, will end up with 1 oz of copper on the outer layers.

Dielectric Selection

Between the layers of copper foil is a dielectric material with specific material properties.  The IPC-4101 standard details the various property requirements of these dielectrics and engineers often refer to slash-sheets, an appendix to the standard.  For your first projects, you'll likely only ever use FR-4 (Fire-Retardant Level 4) materials -- they are the cheapest and most ubiquitous dielectrics in use today.  But if you are designing an RF circuit, you should work with your design-house to investigate more suitable materials.

3.2. Drilling and Routing

Panels are placed in a very accurate drill/mill machine where the through-holes are drilled and the basic board shape is routed.  These machines are accurate to approximately one ten-thousandth of an inch.  A ten-thousandth of an inch is roughly 30 times smaller than a human hair.  The precision is not necessarily needed for modern printed circuit boards, but without the rigidity and precision that these machines offer, shear stresses caused by lateral vibration would cause the tiny drills to break. 

Here's an important note -- even though the machines are accurate to within a ten-thousandths of an inch, the boards will not be located inside the machines to that level of precision.  It's not uncommon to see manufacturers guarantee that holes are drilled within 3 mils of the target location for standard processes and 1 mil for extended processes. 

Hard Way Hughes"If you've ever worked in a machine shop, you'll likely know why the manufacturers only guarantee 3 mils.  Dialing something in with 3 mils over 24 inches isn't impossible, it just takes extra time.  For non-machinists, imagine dropping a thin string twenty-feet straight down from the highest point of your roof to the ground below.  A deviation of 3 mils over 2 feet becomes 30 mils over 20 feet -- in short, any deviation would likely be confined within the diameter of your test string -- you wouldn't be able to tell."

Lenz Mill Drill machine

This LENZ mill/drill machine can handle two boards at once.  The drilling and routing takes place in a single machine, saving operator time and customer money.

The smallest hole that can be drilled with these machines is around 5.9 mils. The limitation is due to the fact that these miniature drills lack the rigidity and strength to punch through copper foil layers -- they tend to bend and deflect (wander) as they encounter copper on various layers of the board.  Additionally, to reach proper cutting speed with such a small diameter, the spindle reaches around 140,000 RPM.  Smaller diameter drills would require even higher spindle speeds.  

5.9 Mil twist drills

These are Royal Circuit Solutions' smallest twist drills ~ 5.9 mils in diameter.

Drill life is calculated, tallied with each job, and each drill is monitored with machine-vision systems.  When a drill has reached the end of its useful life, it is unceremoniously discarded into a five-gallon bucket for recycling.

Twist Drills in Bucket

Used drills are not resharpened -- they are collected and recycled.

It is possible to drill smaller holes in PCBs with laser drill machines, but that will increase the cost of your PCB.

3.3. Electroplating

To prevent copper from going onto areas of the board that will later be etched, the panels go through a photo-sensitive lamination process prior to electroplating.

Hot Roll Laminator

A dry-film laminate is placed on either side of the PCB and then light patterns are projected onto the laminate.  The light polymerizes the plastic when it strikes the laminate and that becomes a protected area.  In the spots where the light didn't strike the laminate, the coating will wash away easily, leaving exposed copper.  

Via Formation Proces

This series of images shows an abridged via-formation by electroplating process.

To form VIAs and thicken traces, PCBs are placed in a chemical bath and connected to the cathode of a power supply. Sacrificial copper ingots are connected to anodes of the same power supply and lowered into the tank with the PCB.  Copper ions travel through the liquid to the panel, where they fill the via cavities and increase the overall thickness of the copper layers.  

Ingot in Bag

Copper Ingots are hung from conductive frames that are connected to the anode of a power supply.

Wet Process Tank

Circuit Board panels are connected to conductive frames that are hung on metal bars connected to the cathode of a power supply.

Board out of Eletroplating

Image of a worker holding a PCB Panel in a conductive frame prior to submersion in etching tank.

This process is used to make through-hole vias.  To make buried or blind vias, additional layers would be added to the PCB after the through-hole vias were made.  So buried and blind vias add additional lamination steps to a PCB design.

3.4. Etching

Etching removes copper from a PCB.  Heavy copper must spend more time in the etching tanks

In the etching process, both sides of the PCB are laminated with a special plastic material that polymerizes when exposed to light. Digital projectors shine light in a pattern that is unique to each copper layer from the design file. This light causes the plastic to toughen. Parts of the plastic that were not exposed to light will wash away in a chemical bath.

The exposed copper dissolves from the board in an acid bath.

Etchback Compensation

Many engineers do not realize this, but the current state[1] of manufacturing does not allow the copper to etch vertically down -- it undercuts at a slight angle.  So to ensure a minimum trace width where the copper meets the dielectric, fabricators have to make all of your copper traces a bit bigger than you asked for.  This is referred to as etchback compensation and it is the reason for the minimum spaces in trace and space charts.  There has to be enough room between the traces after etching compensation to allow the acid to get in and dissolve the copper.  

Trace and Space charts

This chart of preferred minimums at Royal Circuit Solutions.  These minimums can be exceeded at the expense of decreasing yields.

You'll notice that the minimum space for outer and inner traces differ by several mils.  Manufacturers can exceed these minimums if asked, but it will increase the cost of manufacture.  Going beyond these limits is not impossible, it is just a bit risky as it can reduce overall yields.


[1]  Some production methods allow the etching of copper with high-powered lasers that vaporize the copper layers.  Most production lines still rely on acid-baths to remove the copper.

3.5. Lamination

To create a 4-layer, 6-layer, or 32-layer PCB, fabricators separate cores with a material referred to as prepregPrepreg is short for pre-impregnated, and it refers to a dielectric material that is saturated with a thermally-activated epoxy.

Core Prepreg Core

When the cores and prepregs are placed in a hot-oil press for several hours, the epoxy permanently bonds the cores together.

Hot Oil Press
This hot oil press permanently bonds the various layers of a PCB together.

If your board has blind or buried vias, this will increase the number of press operations required to fabricate your PCB.  Each additional press operation adds additional time and increases the risk of something going wrong during fabrication.  So to keep the cost of your PCB down, try to limit the number of press operations required to make your PCB.

3.6. LPI Solder Mask

Liquid Photo-Imagable Solder Mask is the layer of a circuitboard that lies between the outermost copper layers and the parts.  Historically it's often been a green film mask with white silkscreen on top.

Assembled Circuit Boards

This circuitboard appears green because of the LPI Mask layer.  Image courtesy Wikipedia.com

LPI Mask is applied as a liquid that covers the entire board.  Then ultraviolet light is shown on areas of the board where the LPI Mask needs to remain.  A cleaning step dissolves the mask on areas that were not hit by UV light revealing lands, testpoints, and other areas of exposed copper.

LPI serves two practical purposes.  First, it protects copper traces and vias from corrosion and oxidation.  Second, it acts as a solder dam.  Without a LPI Mask, solder would bridge traces and lands creating short circuits and leaving an assembled circuitboard full of short circuits and areas of too-little solder.

3.7. Silkscreen

The silkscreen layer is used for assembly, identification, and component identification for service/maintenance.

The silkscreen layer sits on the outermost layer of a PCB and provides components reference designators as well as pin 1 or anode/cathode identification for polarized parts.

Plenty of companies still apply the silkscreen layer using silkscreen processes, complete with screen, squeegee and pattern templates.

Silkscreen Process from http://www.gwent.org/gem_thick_film.html

Image of silkscreen process from http://www.gwent.org/gem_thick_film.html

But silkscreen printers that utilize ink-jet technology are gaining a larger foothold in the industry.  These devices print directly onto PCBs in the same manner inkjet printers print directly onto a piece of paper.

Inkjet Silkscreen Printer

Image of silkscreen printer from http://www.zhengyekeji.com/

Here's an example of the quality from one of these machines:

Print Sample

Image of silkscreen print quality from http://www.shengyekeji.com/

Hard Way Hughes"The Silkscreen layer is not always aligned as it should be on top of a PCB.  And if your silkscreen print ends up on top of a pad, it will disrupt the solder process.  So don't squeeze things in too close to lands."

4. What is in a microcontroller package?

Most engineers who design circuit boards today have to include at least one integrated circuit in their design.  Decades ago, engineers had to make everything from base components -- transistors, relays, resistors, capacitors, etc...  In the 1960s, manufacturers began perfecting the techniques required to miniaturize those base components into some of the first prefabricated and ready to use packages they called Integrated Circuits (IC).  You likely have experience working with some of these devices in a digital logic class: the 555 timer IC, the 7400 series of integrated circuits, etc...  Nowadays, many of those purpose-built devices have been replaced by programmable microcontrollers such as the ATMega328P we will use in this design.

These integrated circuits are made by carefully exposing incredibly pure silicon to a variety of impurities in a very controlled environment.  Through the application of light, chemicals, vacuum, and heat, these engineers create microscopic transistors, diodes, and other rudimentary elements that make up a microchip or IC.  

This image of a silicon die shows a microcontroller complete with a microprocessor, and areas of RAM and ROM. (Wikipedia)

These silicon dies are often very small, very fragile, and usually difficult to work with directly so they are usually attached to stronger metal lead frames with tiny gold bonding wires.  Then the die and the lead frame are hermetically sealed into an epoxy package and stored in a warehouse at a distribution center until they are needed.  The packaging helps to protect the IC die during storage and use, and it allows for quick testing and installation.

This transparent picture shows a SOIC-14 package interior

There are other packaging options, including BGA and flip-chip packages.  And indeed, the silicon dies can be attached directly to a PCB in a Chip-on-Board configuration.

BGA Package

This artistic impression is an exploded image of a fictional BGA device.

Hard Way Hughes"Why would an electrical engineer ever need to know what's inside a package, just throw them on the board and go, right?  Not really, no.  There is a phenomenon known as 'ground bounce' that is caused by changing logic-levels and the inductive loop formed by the bond wires connecting the leadframe to the die.  'Bypass', also known as 'decoupling' capacitors are used to provide a short burst of electrical energy to the die to prevent the reference potential from changing as well as to localize the phenomena to the chip.  They have to be placed as close as possible to an IC."

5. What is Inside a Passive Package?

Inductors, Resistors, Capacitors, etc... aren't quite as difficult to make as integrated circuits.  The materials that make the component are encapsulated and formed into standard-sized packages with two or more metal pads that are later soldered to a printed circuit board.

Component Value Calculations

While electrical engineers would like to have every possible value of every component made available to them, this is simply not practical from a manufacturing and inventory management standpoint.  So manufacturers provide standard component values for each order-of-magnitude (decade) based mostly on the equation

E Series Approximation

where m is the E-Series value (6, 12, 24, 48, 96, 192) and n is a positive integer less than m.

Each order-of-magnitude is logarithmically broken into a set number of values.  For E6 series, those values are 1, 1.5, 2.2, 3.3, 4.7, 6.8E6-resistors would have values of

\sqrt[6]{10^0}=1 \sqrt[6]{10^1}\approx 1.5 \sqrt[6]{10^2}\approx 2.2 \sqrt[6]{10^3}\approx 3.3 \sqrt[6]{10^4}\approx 4.7 \sqrt[6]{10^5}\approx 6.8
1 Ω 1.5 Ω 2.2 Ω 3.3 Ω 4.7 Ω 6.8 Ω
10 Ω 15 Ω 22 Ω 33 Ω 47 Ω 68 Ω
100 Ω 150 Ω 220 Ω 330 Ω 470 Ω 680 Ω
1 kΩ 1.5 kΩ 2.2 kΩ 3.3 kΩ 4.7 kΩ 6.8 kΩ

and the series would repeat for each multiple of 10 up into the MΩ range.

There are E-Series divisions into 6, 12, 24, 48, 96, and 192 pieces.  Each series provide additional values that were not seen in the series beneath it.  And very tight-tolerance resistors are available for high E-series devices.

E Series Numberline Plot

As the number of values in the series increases, the tolerance tends to improve, and the cost of the part tends to increase.

If you create a design that requires a 481 Ω 0.1% resistor, take note that 481 Ω only exists as a value in the E192 series -- which means it's going to be significantly more expensive than a 470 Ω 1% resistor taken from the E24 series.

481 Chip Resistor

This is a screenshot of a 481-Ohm E192 series resistor from Digikey.com on 2/17/2020.  At 1kU prices, the components are still $0.12 per part.

470 Ohm Resistor

This is a screenshot of a 470-Ohm E24 series resistor from Digikey.com on 2/17/2020.  At 1kU prices, these parts are less than $0.01.  As you can see, you can spend less to purchase 100,000 of these resistors than 1000 of the 481 Ω resistor shown above

If you need non-standard values of resistors, create them using a series/parallel combination of standard resistors.  Be certain to keep track of uncertainty of measurement as well as power dissipation factors when doing it.

Hard Way Hughes

"You can, I'm certain, order a 500-unit run of 482.3 Ω, laser-trimmed resistors for your next design from a variety of manufacturers.  Just make sure your resume is up to date before submitting the request through purchasing.  The rarer an item, the more expensive it is going to be.

You should also know that I can count on one finger the number of conversations I've had with other engineers about 'E-series' components -- it's generally only discussed by component engineers.  Just be aware that there's a reason some components cost more than others.  Your projects will tend to be more successful and your career will generally last longer if you actively choose the cheaper components wherever possible.  Save the pricier parts for high-end designs."

6. Putting it all Together

All parts are mechanically, electrically, and thermally connected to printed circuit boards with solder.  Solder is a combination of metals that are chosen for their mechanical and chemical properties as well as their price and environmental impact.

Periodic Table with Highlighted Metals

This periodic table highlights common elements used in solder-metals.

But the metals aren't mixed together randomly.  They are mixed in exact proportions that provide specific thermal and mechanical properties.

Solder Alloys

Solder compositions made of different metals have different transition temperatures.

Hard Way Hughes"Engineers rarely ever consider the solder that is used to connect their parts to their PCBs -- until bad things start happening in their designs.  Then there's a pretty intense study session and a lot of phone calls to other designers to try to figure out what is going on.  There are many solders to choose from, and even something called Transient Liquid Phase Sintering Paste -- which will only melt once."

6.1. Electrical and Thermal Conductivity of Solder

Materials that conduct electricity well, often conduct heat well.  Of the common PCB materials, Copper offers the best compromise between price, conductivity, and manufacturability -- and it happens to be the second-best electrical and thermal conductor available to PCB manufacturers.  Copper has a greater conductivity than aluminum and even gold.

Thermal and Electrical Conductivity of Common PCB Elements.

This graph shows the relationship between thermal and electrical conductivity for common PCB metals.

Copper is used on PCBs to simultaneously conduct electricity and heat.

6.2. Eutectic Compositions

Solder is a type of metal chosen for its electrical, mechanical, and thermodynamic properties.  It is generally composed of two or more metals.  Historically, those metals have been tin (Sn) and lead (Pb).  When tin and lead are mixed together, they form an alloy whose melting point is lower than the melting point of either individual metal. 

Most alloy concentrations of lead and tin have an intermediate stage of matter that appears to be a type of sludge or paste -- neither completely solid nor completely liquid.  But when the two elements are mixed together in an exact proportion of 63% Tin and 37% Lead (Sn63Pb37), the metals transition directly from solid to liquid and from liquid to solid at a lower temperature and without the intermediate states-of-matter ever appearing.  This precise mixture transitions at a single, low temperature and the alloy is referred to as an Eutectic composition.

Eutectic Compositions

This is a common Eutectic Diagram that shows the mass percentage of Lead and Tin on the horizontal axis and the phase transition temperatures on the vertical axis.

Not all solder compositions are eutectic -- but they all do try to minimize the transition temperatures into and out of the sludge/paste phase.  

Solder Alloys

Assorted Solder Compositions are shown in the table above.

The theory for the rapid transition is that it can keep contaminants from congealing into large clumps during the cool-down phase.  Large clumps of contaminants become mechanically weak areas where a solder-joint might break.

6.3. Intermetallic Bonds

The flux is solder-paste melts at a lower temperature than the metals.  Liquid flux flows over exposed base metals, removing the oxides and surface contaminants and providing a temporary barrier against oxidation.  Additionally, flux acts to lower the surface tension of liquid solder metal -- allowing it to spread further along flat surfaces and draw higher to form stronger fillets on vertical surfaces.  

Solder Reaction Layers

Flux helps to both protect the reaction layer from oxidation and to help extend its reach along a base metal.

When the solder-paste reaches a slightly higher temperature, solder-metals liquefy and begin to displace the flux as they move over the metallic surfaces.  While in this liquid form, the tin in the solder metals begins to form intermetallic bonds where it interfaces with copper pads, pins, and lands.  These intermetallics are what mechanically attach a pad/pin to solder, and solder to a land.

Intermetallics are crystalline arrangements of molecules that behave similar to pure metals.  Unlike an alloy, where elements are simply well mixed, intermetallics exist in exact stochiometric proportions.  For example: Cu6Sn5 and Cu3Sn.

Intermetallic Bonds

These example crystals show some of the different shapes molecules might arrange themselves in to form a crystal.

Intermetallics

doktori.bme.hu

This scanning electron microscopy cross-section shows the mixture of tin and lead solder in an alloy and an intermetallic layer.

It's important not to allow the solder to remain liquid for too long, as large intermetallic crystals will form, and these crystals become points of weakness in a solder joint.

SolderJointFailure

This solder joint is in the process of failing due to excessive intermetallic formation, likely caused by prolonged melt-time during assembly.