Electronics Construction Techniques

Brooke Clarke, N6GCE


< 1955 Point to Point Wiring
1955 Terminal Board Wiring
Modern Terminal Board
1960 Single Sided Printed Circuit Boards
>1960 Double Sided PCB
1970 Multilayer PCB
1970 Surface Mount Technology (SMT)
Throwaway Products (2000+/-)
1990 On Line PCB houses
Hints and Tips or What Goes Wrong - weperate web page

    The dates shown are just a guess.  If you have any date info let me know.  The dates are just guidelines.  The technologies overlap.  For example in 2006 both through hole and surface mount is being done.  But there is a trend in the IC houses to offer only SMD packages on most of the new parts.  At some time in the future through hole ICs and other through hole components will be looked at like tubes are now.

Point to Point Wiring - <1955?

In this method the larger components are mounted on a chassis and controls are mounted on the front panel. Smaller components are connected directly between those already mounted.  Where an intermediate tie point is required a terminal strip is mounted to the chassis.  A terminal strip is a piece of phenolic with one or more terminals pointing up and all insulated from each other.  There's at least one attachment point which has a terminal at the top for a chassis ground connection and at the bottom there's a foot with a screw hole so that terminal strip can be attached to the chassis.
In this technique the assembly line is linear, that is to say all the mechanical work, paint and labeling must come before any electrical assembly.

A "bread board" was literally made using a wooden board used for kneading bread dough.

This was the method used for all electrical construction up to the 1950 time period.  The active devices were vacuum tubes.  The filament power of a tube needs to be about the same as the output power rating so that tube devices always have an energy efficiency of less than 50%, sometimes way less.  I knew a man who worked for RCA and his job was to design color TV picture tubes so that they would fail just after one year (the common warranty period).  I think that vacuum tubes were also designed to have a 1 year operational life so that you would buy more tubes.  Every drug store had a tube tester with the common consumer part numbers in the base.  I made many trips there with a paper bag filled with tubes, one of which would be found dead.

A common pilot lamp setup was to use a faceted glass window through the front panel mounted to a socket which held a number 47 lamp.  This was powered by the filament winding of the power transformer. When lit you knew that the line fuse was OK and the transformer was putting out filament voltage.  Another option was to use a Neon lamp, but these were not as bright and could not be filtered to give many different colors, but they did not burn out like the filament style lamps.

If you connect a 90 Volt battery in series with a resistor and then connect a Neon lamp and capacitor to ground you get a relaxation oscillator.  The RC circuit charges up and when the voltage gets to the Neon firing voltage the Neon lights and pulls charge from the capacitor.  Once the voltage gets below some value the Neon lamp goes out and you wait for the next charge time.  The effect is to have a lamp blinking all the time and mine was about once per second.

Radio receivers built in this time frame did not have enough frequency accuracy to tune to a station, you needed to tune above and below the indicated frequency until the station was found.  You could not just tune to the correct frequency and wait to hear it.  Once the station was heard you could record the logging dial setting which typically was a number between 0 and 100 on the band spread dial.

Terminal Board Wiring - 1955

This is an extension of the terminal strip.  It consists of an insulating board (Masonite, phenolic, etc.) about 2 inches high and maybe 6 inches long (varies with what is on it).  There are terminals along both long edges with components mounted between the terminals and one or more wire cables leaving that are pre shaped and cut to length.  This board can be assembled and tested prior to any chassis hardware being available, shortening the span time on the assembly line.

This was used for vacuum tube and some germanium transistor equipment.  Early HP equipment used terminal boards.

Modern Terminal Board

I first saw this at work used by a VHF-UHF expert.  He started out with a blank double sided printed circuit board sized to fit a specific box with mounting holes.  The terminals were of two types, both were small teflon cylinders with the terminal on the center line.  The teflon has a shoulder so that when the terminal is pressed into a hole in the PCb the shoulder stops it form going all the way through.  One type was a single sided terminal the other type had a terminal on both sides.  The later allowed DC circuitry to be on the bottom of the PCB and the single terminal type was for the VHF UHF circuitry on the top side.  This would also be good for circuits that were high impedance like low power circuits where printed circuit board parasitic capacitance and/or leakage is undesirable.

Single Sided Printed Circuit Board - 1960?

This is an extension of the terminal board where a  layer of Copper is added to the terminal board that can be etched to replace the component to component wiring and the terminals can be replaced with holes for the component leads "through hole construction".  A number of the components need to have new form factors for direct mounting to the PCB.  Television sets made using tubes and PCBs had failures when the heat caused the cheap  FR4 FR2, similar to NEMA grade XXXP (MIL-P-3115) boards to crack opening traces.

Soldering all the connections after the board is stuffed is time consuming and error prone.  Wave soldering was developed to solder all the joints in a very short time.  A common costing method was to count the number of holes in the board.  This lead to more functions in a circuit package to minimize the number of pins.

Silicon transistors and early integrated circuits (IC) used this method.  The pin spacing on early Dual Inline Plastic (DIP) ICs was 0.1".  This made for a standardized 0.1" grid for PCB layout.

When it was desired to connect a test instrument to a PDP-11 core memory computer, or other device for remote control, connectors were added to the back with a wire for each switch contact on the front panel or each "bit" of data.  This required connectors with large numbers of contacts and a customized card in the computer for each instrument.  One of the control positions would correspond to all open switches or a separate local-remote switch was used to lift the common switch connections thus allowing the remote to function independent of the mechanical switch positions.  When being remotely controlled these instruments typically would not show any indication on the front panel of what was going on.

HP developed HP-IB (later IEEE-488, later GP-IB) as a standardized way to remote control instruments.  It has a handshake scheme so that the slowest instrument gets it's data before the next parallel 8 bit byte is sent.  There are also a number of dedicated control lines.

Double Sided Printed Circuit Board - late 1960s?

This is a simple extension from single sided PCB but requires a way to connect traces on opposite sides of the board.  I worked with some early DS PCBs and we used a shoe eyelet machine to install eyelets (like very small rivets except they have a hole through the center).
Later a reliable "via" was developed so that a PCB board shop could plate through the via holes.

A version of this is called "cord wood" where instead of mounting components to a single board they are mounted between two boards.  This results in a very dense packing of the components and allows non planar shapes, like a hollow cylinder to be made.  The RS-49 was the first CIA made radio to use this method and that was in the 1963 - 1964 time frame (they called it "sandwiched").  The GRA-71 keyer uses cord wood.

Another version of this is the flex circuit.  Instead of using FR4 or fiberglass for the board material a thin flexable plastic is used.  This is great for things like wiring the controls on the panel of a PRC-68 radio.

Yet another version is "Chip On Board" (COB) where the semiconductor die is attached directly to the board.  It could be with the bonding pads up and conventional wire bonding used to connect between the board and the chip.  It could be a bumped chip where there is a solder drop on each bonding pad and the chip is soldered to a flex circuit.  The flex circuit and chip are cut out and the flex circuit attached into a package or to a board.  This is used for things like greating cards and other low cost high volume applications.

Bread boarding and small run prototyping could be done using wire wrap technology.  The component side of one of many PCBs lookes like this Stanford Telecommunications 5001 GPS simulator board.  This instrument is serial number 4.

Another option was to make a PCB by making a 4X trace layout using Mylar grid paper and black tape with special "donuts" for the component holes and have life size a litho negative made.  The PCB shop would use this to etch the board.  Double sided boards required alignment marks on each negative done is such a way that they would not be used flipped or reversed.

This method was used for Silicon transistors and was very popular for ICs.  Multifunction ICs and early Application Specific Integrated Circuits (ASIC) used this method.  A PCB with hundreds of general purpose ICs could be replaced by a signle board with just a few ASICs greatly reducing the number of holes and therefore the cost.  Since ASICs are not general purpose, repairs become harder if not impossible.

Nuvistors were tubes made using ceramic cylinders instead of the conventional vacuum tube glass bulbs.  They could run on lower voltages and were promoted as an alternative to transisitors by some tube manufacturers.

Remote control was by means of RS-232 or HP-IB (IEEE-488, GP-IB) interface.  The controls were no longer mechanical point to point but instead momentary contact switches (normally open) with some type of indicator to show the state of each switch.

HP developed 10Base-T, a way to use existing premises wiring for computer networking as a much lower cost alternative to installing coax cables.  This developed into 100Base-T (CAT5) and higher speed versions that need special Unshielded Twisted Pair (UTP) cabling.

Many things were made using general purpose parts that could be repaired.  These are in my opinion the "Golden Era" products because they can be maintained almost forever.  When a newer design fails the only option is to get replacement modules form the factory, and if the factory is no longer selling them it's very difficult or impossible.  There are a few parts in the PRC-68 series radios that are ASICs or programmed microcontrollers that can not be purchased new now, but most of the rest of the radios are made from general purpose parts and are repairable.

Some PRC-68 radios were made using two double sided boards separated by 1/16".  At that time miltilayer boards were not available.

The Austron 2100F  LORAN-C Frequency Monitor and 2100T LORAN-C Timing Receiver use double sided Printed Circuit Boards and have documented HP 5004 Signature Analyzer data to aid in troubleshooting to the IC level.  This is probably the last of the equipment designed so that individual ICs can be replaced.  After this troubleshooting is done by the board swap method or even equipment swap, like for a modern TV set.

Multilayer Printed Circuit Board - 1970s?

Once a via can be made, it's a natural extension to make a PCB with multiple layers which is how computers and other complex electronics boards are made now.

Complete function on a chip packages are used with this method as well as all of the above.  It is impossible to repair a multiple layer board and even replacing a component can be extreamly difficult or impossible depending on the packaging method.  Most of these are replace instead of repair.

Vacuum tube technology developed so that a tube would last as long as a transistor.  The PRC-25 Viet Nam vintage radio has a final 2D24 tube thats known to never fail under normal conditions.

Later versions of the PRC-68 use a single multilayer PCB instead of the two seperate baords.

ExpressPCB offers 4 layer boards where the center two layers are always power and ground planes.  The top and bottom layers are normal user wiring.  The capactance between the power and gound planes is a good thing.  Also both power and ground planes are connected by capacitors so both are good RF grounds which is a good thing if you need to meet FCC radiation standards.  Whenever there's a hole from the top to the bottom that should not connect to either power or gorund, the center layers have a hole in the metal patters so that the via will not make electrical connection to the center layers.  It's much easier to layout a 4 layer board since there's no power traces to work around.

Surface Mount Technology (SMT)- 1970s

Rather than drill holes in the board, the parts are mounted on the surface.  This greatly reduces lead lengths thus extending the upper frequency of operation.  The assembly method is to use an automated machine to put a dab of glue under where each part will end up, then use a pick and place machine to put each SMT part on the board, the use a production line soldering method like vapor phase reflow to make all the solder joints.  Some thorough hole parts and be automatically placed, but then may need hand soldering.  Parts can be placed on both sides of a board.  This is the lowest cost method as of 2004 for making boards in volume.

Throwaway Products (2000+/-)

Prior to W.W. II watches were hand made and included jewels as bearings.  They were purchased from a Jewelry store that could clean and regulate the movement.  When bomb fuses were made into high volume products the same production equipment was later used to make Timex watches which had no jeweled bearings and were not serviceable, so jewelers would not carry them.  Instead they were sold by drug and other discount stores.
It takes a licking and keeps on ticking (YouTube)

A similar thing has happened to consumer (and other) electronic equipment.  It used to be that a TV or radio was repairable and sold by a store that could maintain it.  Now consumer electronic equipment is sold by discount stores and the cost is so low that it is a throwaway item.

2006 - the local paper is advertising new VCRs for $25 and DVD players for $35.

On Line PCB houses

There are a number of on line PCB fabrication houses, like ExpressPCB, that have a pricing structure based on:

PCB Software

Professional PCB design is done using a layout program that has output files in Gerber format for the traces and also a drill file.  The advantage of this is that it's a universal format that's accepted by a number of PCB manufacturing houses.  The disadvantage is that it's a lot more complicated than the free  software provided by ExpressPCB.

The Express PCB software is in two parts.  The old one is just the mechanical board layout software.  The problem with using just the layout is that you may or may not connect all the pads needed for each node.  It seemed that on the simple boards I was making there would always be either a missing trace or a trace shorting out something.  Now they have a schematic capture program (which can be used all by itself) that can be linked by the layout program.  This makes it very easy to confirm each network node is properly connected.

Note that they do NOT count the number of holes for standard pricing (except there is a maximum number for any board, and if exceeded they either will not make the board or charge extra for each hole in excess).  I found this out after designing a "prototype" board that had a lot of holes, and when trying to use a number of small holes to make separating a panel easier.)  The holes are drilled by a robot and the machine time is not that much of a factor for a normal board.  If you had them make a board with a grid of holes there will be a surcharge.
If you have introduction dates or more on this topic let me know brooke@pacific.net.

Back to Brooke's Hints & Tips (or What goes wrong)PRC68.COM, Military Information, Electronics, Home page

[an error occurred while processing this directive] page created 6 July 2001.