Computer Aided Drafting (CAD) of Printed Wiring Boards A rambling discourse giving background information and the history of electronic printed wiring board artwork design from the 1940's to the present day, based on my own experiences in Australia, the USA and other countries. by Bob Barnes In the beginning I (and others) used to draw up early printed board artworks at eight times actual size, using a stencil made from a sheet of perspex about 3mm thick (1/8"), with milled out holes eight times the size of the pads we wanted. This stencil would be placed on the drawing paper with the centre of the 'pad hole' exactly over where a pad was to go, and then a small stiff brush was used to apply Indian ink to paper through the stencil hole. Then the stencil would be lifted and moved to the next pad position (say 8/10" away, when you were doing IC pads) before repeating the process. It was a very slow and laborious process, but very exact, considering the fact that you were working at 800% of the final size. In any case there was NO other way,-- this was it, the real start of board artwork drawing. The drawing was done either on A3 sheets of white paper jointed with sticky tape (for 'reflective' photography) or similar A3 sheets of frosty velumn (paper thin frosty film), again stuck together with sticky tape (for 'transmissive' photography, as it was called then). All this work was done before the invention of photocopiers as we know them today which commonly using A4 80gsm bond. We used to have a two-part copying machine made by 3M, but using it for PCB artwork was folly and never successful, very much thermaly stretchable (totally non-linear) during the play out process. After the 8x single sided paper masterpiece 'artwork' was drawn up, you would then take it into a large windowless room painted matt black, which in one wall housed a 'Schneider Lens' (about 1.5m up from the floor). Your artwork was fixed on the wall opposite the lens, and there were a couple of arc lights on either side aiming at the artwork through about 45 degrees, to light it evenly and not permit 'bounce' glare. On the other side of the lens was another smaller room, lit only by 2 x 25w lamps fitted inside RED filtered safelights (RUBYLITH to the rescue) This room is virtually the insides of a very large 'box brownie' camera, and inside this, there was a simple low pressure vacuum frame (about 20 x 24 inches) which could hold a single sheet of Ortho Type 3 photographic film. The vacuum frame was usually a good quality 2000w barrel vacuum cleaner This very large camera system was used to make an 8:1 reduction copy of your original artwork. (or 2:1 for further minor 'mods'), Both the camera and its vacuum frame were taken from the printing industry, where they were used to make printing plates for offset or pink stereo's for letterpress printing presses. When the sheet of film had been placed in the vacuum frame for exposure the arc lights on the main room were turned on to illuminate your paper artwork masterpiece (or a bank of 40w fluorescent lamps turned on behind your transmissive artwork). All this can be done automatically with the shutter timer mech, but first (before Ortho Film Loading) you operated the mechanical shutter mechanism in the Schneider lens, keeping it open for about 20 secs at approx f/22 if I remember rightly. This permitted the image of your artwork (inverted) to appear on the film. (be it Red, Black) Many 'vacuum' frames were a simple vacumm cleaner providing a less air pressure surface on a perforated peg board, and the film 'stuck' Oh, before we did the actual exposure, we had to do a quick check to make sure the image was focussed and in the centre of the film work area, using a sheet of thick frosty velumn placed in the film holder as a pre 'view screen'. Then we turn off the RED safe light lighting, turn on the artwork image lighting, check all is ok for 'centre', reset the lens shutter, turned the safe lights back on, put the Ortho Type 3 film into the camera film holder, close the back, then hit the 20-second timer button which turns ON the artwork image illumination. When the shutter finally closed again it would automatically turn OFF the artwork image lights and the film vacuum as well as turn back on the red safelights, so we could take down the film and develop it in a tray of Ilford or Kodak developer. After development the film looked like a mostly black area with white pads and tracks on a very high contrast negative of the image on your artwork on paper or translucent velumn text were on the original 8x artwork. If this negative image looked good, we gave it a quick flash rinse in some older photographic used 'fixer' solution, which was used as an el-cheapo 'stop bath'. Then it was placed in a tray of reasonably fresh Kodak or Ilford Fixer solution, where it would stay for about 3 to 5 minutes. This made the 'white' pads and tracks go clear, by dissolving away the unexposed silver. The process was then basically complete, so the negative film was given a quick wash in a 'soapy' solution for 5 minutes, then a light rinse in fresh water. After this it was hung up to drip dry in the air, ready for the next step in the board making process -- using it to expose a piece of FR-3 or FR-4 (or whatever) copper laminated epoxy fibreglass which had been coated with Kodak KAR or KPR photo-resist. I'll discuss this next step later, but before that I want to talk about the way we used to produce the basic board artwork a few years after the Indian ink-brush-stencil and paper or velumn approach I've just described. This was using the adhesive-backed 'puppets' or 'donuts and tracks' which were made by a firm called Bishop Graphics USA, and sold in Australia by a company in Bexley, NSW called Circuit Components -- run by a patient and absolute gentleman, Douglas ('Doug') Rees. Making your artwork this way was done on a light table, a table with a smooth frosted glass surface, lit from underneath by an array/bank of fluorescent tubes. The artwork was now done on frosty velumn with a 1/10" (100 mil or thou) faint blue grid on it, so when the velumn was taped to the top of the light table it provided an accurate guide for placing the Pad track strips and 'D70 etc donuts' As a quick aside, I should explain why a grid based on Imperial measurements (25, 50 or 100 mil spacing) was adopted, and is still the world wide standard. Partly this was because virtually all ICs and most other components had (and still have) their pins and leads spaced on these multiples, so it made sense to use 25/50/100 (thou) mil spacing for every thing else on the boards. Another reason was that when a CNC paper tape was used for drilling the holes in your boards, this would be a lot cheaper to make (and faster to use) if all holes were spaced in these multiples. This ensured that there were no silly X and Y end points in oddball numbers of mils/thou, like 53 or 87, in which CAD location could be 10053 or 10087 for a dimension location of a 'pad' at the 10 inch x 10 inch placement. This is also why we board manufacturers don't like getting artworks done to a metric grid, like they produce in some European countries. CNC drilling machines work in mils / thou, but they are much slower if they have to operate in 1 thou steps in order to drill a board with the pads laid out on a metric grid, sure it can be done, but it is slower drill for production. One company here in Australia closed some years ago because they accepted a board design from a Scandinavian country, after no one in Europe would touch it with a barge pole and it had been drawn on a fully metric 1mm layout. The order was for a very large number of boards, so CNC drilling had to be used. But instead of the CNC machine being able to scoot around in 25thou steps, it had to operate in 1 thou steps - which resulted in it ramp up/ramp down travelling in 1 thou increments, thus taking much longer. Because the penalty 'late delivery clauses' would have bankrupted the company, so, they closed before legal action could have eventuated. While it may seem that the X-Y location of a hole is not that important, the fact is that every move of the CNC drill head involves ramping up speed, then ramping down and stopping. If this all has to be done in 'hi- res' 1 thou mode, it does causes serious time loss in the automated mass production of circuit boards. By the way, the CORRECT terminology for a circuit board is a 'PRINTED WIRING BOARD', not as most less 'aware' people call them 'PRINTED CIRCUIT BOARDS'. True, printed circuits are usually done on a printing press, be- it a Heidelberg KOR, GTO, Platen etc, like your A4 letterheads are done these days for a business. But components (like resistors, capacitors and inductors) as well as the wiring was printed (in black/white) directly on the paper material using common black (or coloured) inks. In contrast, printed wiring boards were originally designed to replace the old technique of a wiring loom, where all of the wires (usually cotton- covered) used to make the connections in a car, plane or tank were precut to length and laced together to form a complete assembly, ready to install in the vehicle or equipment. In fact one of the first uses for them were as 'printed wiring looms' like the one still in your motor car, up behind the dashboard. Who invented the process? The process of making printed wiring boards was actually 'invented' during the second world war (late 1942-early 1943) by a chap in England named Paul Eisler. He later wrote a book about it ('My Life With the Printed Circuit", ISBN 0-934223-04-1), and it's a time capsule of fact copied by many others. I seem to remember he was of European extraction, having fled from Nazi Germany. It was his 'side kick' Sean McGovern who developed the new types of 'acid proof inks' it needed. When I first met Mr McGovern (about 1972) I found him at Cheltenham, England, a likeable sort of Irish rogue, but nearly impossible to understand with his broad Irish accent. Later during the 1980's I met with Sean again many times, (after he migrated out here to Australia) and it was truly an enviable pleasure to chat with him as his brand of Irish/English was by then changing to 'Iri-Aus' and our conversations will always be 'imprinted' always in my memory. Sean told to me that Motorola in the USA tried to have Congress ban the process of making circuit boards made in the USA for many years. This was because it would destroy the jobs of the huge numbers of lads & lassies in factories who made the cotton covered wiring looms for everything: tanks, automobiles, planes etc especially during WW2.. It would be many many years after WW2 that the USA would get involved with the circuit board industry and be legally permitted to make their own. It's also rather amusing that only last week I had a customer asking about 'blue' blank circuit board raw stock material. I seem to recall that when I bought RCS Radio off the original owner (back a zillion years ago), we used to have some raw 4' x 3' sheets of 'blue' based 'impregnated' fibreglass board material, and guess what, it was proudly made by Motorola ! amazing. Raw Blank board stock material could have been made here in Australia by a company up at Telopea (Sydney) called 'FORMICA' but was banned from doing so because at the time the govt. stated that the making of circuit boards were only a flash in the pan, like a hula hoop, here today and gone tomorrow and those days the with IMPEX dept, you had to get govt. permission to buy any equipment from overseas, balance of trade regs. etc. Here in Australia, the law that required a red flag carrier to walk in front of a motor car on a public road was only removed from the statutes in the last decade or so - so most countries have their weird 'old laws'. If you ever learn the true history as to why Sydney didn't get electric lighting some 16 years after a little country town called Tamworth, you'll come to the conclusion that 'money' is above reality & commonsense also there in not much of a whaling industry in Tamworth, dunno why. The same applies to the situation regarding today's general wire cabling in military aircraft. This is vastly superior in insulation to the wiring in more common jet aircraft for conveying the public, because after all, it takes a lot of time and money to train a military pilot & air photographer. Whereas with the general public, it's 'just another ticket sale'. You learn a new swear word when in the military services, it is 'Next !'. Anyhow, the founder of RCS Radio, Mr Ron Bell was over in England during early 1943 and he was 'invited' to make circuit boards under agreement to Paul Eisler's company which was called 'Telegraph Printed Circuits'. For many years circuit boards made by RCS had code numbers starting with 'TPC' to indicate that they were made under the seven year patent licence, although Paul Eisler gave permission for Ron Bell (several times) to NOT bother with the patent I.D. but he did not do so for quite a while. Sean McGovern told me that when the company's business name was being registered, Paul Eisler could not get the patent office girls to pronounce the word 'Wiring'. So they omitted a word in the business name and instead of calling it 'Telegraph Printed Wiring Circuits' it was shortened to 'Telegraph Printed Circuits', so he was basically stuck with the shortened name either his pronunciation or the office girls misunderstanding Anyhow Mr Ron Bell brought back the process to make 'mercuric timers' for limpet mines (under agreement) during WW2. These were sent back to Coventry in England and used quite successfully by the British Navy. The actual development and manufacture of circuit boards here in Australia was a national secret from 1943 until late 1950, and it took another eight years to get the local magazine 'Radio & Hobbies' to come up with a hobby public project using them for common public use. The first was in 1958. I have no doubt, others (and many have done so over the years, and still do) claim to have been the first to do it, but the British Patent Office recognises Paul Eisler as the 'originator' of printed wiring boards. That is why he held the patent from 1943 until 1950, and I sort of reckon the British Patent Office are very diligent and non-corruptable, so I ignore any others to the claim as pure entertainment and 'pulp fiction' (to borrow a line). You will find various magazines attempting rewrite history to boost sales and their ego, based on misguided rubbish, good on them. O.K. after digressing, I'll return to the CAD overview and continue talking about making board artwork using Bishop Graphics adhesive pads and tapes. Before you sat down with the No.6 Exacto knife to start laying your artwork as Black Pad donuts, puppets of IC layouts and Black Crepe tape, you would have done a quick blue pencil sketch layout as to where you wanted the input/output connections, volume controls and so on. This sketch was also a rough as guts layout of the way you expected to be able to arrange the wiring -- bearing in mind the area available in the intended final housing. This act of pre-drawing the best method of routing new tracks, under components and what seems like around the world for a zac, was an essential step before one sat down with donuts and tape. It was commonly called 'pencilling up the proposed artwork' and using a faint blue pencil as faint blue was camera 'obscura' non-seeable, using current films of the day. In those days when we had miles per hour, feet and inches, we all used 'thou' for accuracy. A thou is 1/1000 of an inch, or what the USA calls a 'mil'. IC pads were spaced by a simple 100 thou, or to the layman they were called 'tenths' or 1/10" inch . So if you are going to start designing artworks for electronics, you need to think in inches, or 1/10" (tenths). Metric would account for about o.01% (if that) of the electronic designs made these days. Many years ago, when I was working at a place in the states that made jet aircraft, I was pulled up and questioned about the new emerging 'metric' size and the new term was a millimetre. Now this was going to be to the masses, Joe Public, an equivalent size known at 2.54mm per tenth of an inch. Well, I quickly learned that in aviation that measurement is NOT regarded as correct; from memory it is 2.5367mm approx. So when you see in future, measurements given in millimetres, bear in mind that they are approximate 'conversions'. Maybe they're near enough for cake making and decorating, but not nearly accurate enough in precise engineering terms for serious electronic equipment, jet turbines, lasers etc. What ever measurement you use, that's it, be it millimeters, or thou or beer cans, a measurement is a measurement standard, and so far I have NOT seen an accurate measurement converter, that you would bet your life on. A glance at T.I. or McMurdo data sheets, so as they can sell to the less CAD experienced locals of Europe, they will show dimensions in metric. Under NO circumstances do you take ANY notice of these dimensions to produce 'footprints' for your CAD software or you will get a 'cumulative' 'compound' error if you ever design a board for CNC placement of misc components, maybe good enough for manual hand insertion, but useless in the real world. Even Japan, China, Korea design mass production in Imperial. The modern 'surface mounted' components, designed these days are done on a 10thou block and multiple parts thererof, however you will see some oddball metric dimensions given like 1.27mm etc, well this is rubbish, and if you tread this way, your mulit-thousand Pick and Place machine for automation will chuck a hernia as that dimension does NOT compute, the true dimension that you will design with is 50thou and if you are silly enough to compond out that measurement into a string of surface mount resistors side by side then the expensive finished design, has been a wasted exercise in money. I suggest you look in the saturday paper for a new job, as the owner of the business, when he gets the invoice/bill from the board maker for junk, it will not enhance your future 'job reference' prospects. However at this tin pot place where they made a 707 Jet aircraft every 23 hours, they have over the years established 'standards' for printed board design. At a time when I was on about $68 per week I would be fined one dollar for every right angled turn I mislaid out, or forgot to put small 'chamfers' so as to prevent problems during etching. The chamfers were also to reduce the risk of mechanical resonance later when the circuit board was in service. In the first week I lost over $30, but after three months I was down to maybe one drawing error a month. Losing dollars out of a meagre wage is an excellent teacher of fixing poor design standards and mistakes. So, from day one, understand inches and thou and do not bother a serious board maker with your artwork laid out on a metric grid. They may not even bother to answer your costing questions. Now, 40+ years later, it's beneath my station to even look at a metric artwork as I know the person who designed it is probably still a cake decorator and has only just looked at electronic design for entertainment/hobby value the night before rather than watch re-runs of Days of Our Lives or Neighbours etc. Anyhow, Bishops Graphics arrived with the pre-slit Black and Red tapes in 40 thou and 100 thou widths (plus others we did not use) for tracks, plus D70 donuts for pads and the 'puppets' for IC layouts (pre-cut drop and plop 14-pad or 16-pad quick templates). Sometimes the artwork was laid out twice final size (2:1), but mostly it was 1:1 -- which was a true skill or craft. Having had the US275 military standards bashed into me and reinforced many times, I modestly considered myself a reasonably talented artwork layer. But I was brought down to earth when I later saw a chappie over in College Street in Gladesville called Terry Flowers. Whereas my (biased) pre-self assessment gave me say an 8 or 9 out of 10, BUT if Terry Flowers was a 10, then I realised that I was really a minus 3. He had the ability to blue pencil up, and work over a large A0 sheet of velumn drafting film like a creeping bacteria on a time delay speeded-up movie. It was truly an example of sheer patience and the ability to pre-plan. I will never forget his ability to work out in his mind the track routing as he progressed -- incredible !, unequalled by only 1 other in 40 years, certainly NOT me. A circuit board will etch about 1 or 2 thou under the edges of tracks and pads during the high pressure acid etching process. And when the components are loaded, the tracks WILL break at any 'right angled' changes in direction during prolonged resonance vibration. Once you have seen high speed photography of this you will NEVER FORGET this visible demonstration. It is also poor design to have component PADS directly on the tracks, because this violates the integrity of the track. A small chamfered spur or stub track is really the only way. Designers who ignore these basic fundamentals of physics will pay the price after the post mortems on a failed circuit board, later on. I recall that a very large electronics company in Japan, back in the early 1970's designed a circuit board with Pads for LEDs spaced at 3mm. The designer was presumably 'arrogant' and not going to use 1 tenth of an inch or (2.54mm whatever). Now the wire legs on a LED are steel, with copper, silver, tin lead or whatever coating. And the steel leads were stressed like a roast chicken's wishbone when they were loaded into the 3mm spaced holes. Eventually the body of the LEDs cracked and failed. This did not amuse the insurance underwriting comp any in the USA called 'UL', with catastrophic results. The Japanese company suffered from poor credibility for many years, and even now they are not entirely trusted. One thing to be aware of is that having an electronics background does NOT ensure the artwork designer knows what he or she is doing in board layout. For example I reckon the best two artwork designers here in Sydney would NOT know a 555 from a 741. Conversely one of the most talented engineers here at an electronics magazine presents board designs that make one shudder and several of his artworks are reduced some 50% in file size by simple 'old fashioned' methods/techniques & principles. Terry Flowers (who may be 'only' a draftsman) would lay out one or three sheets of velumn film, in case the board was going to be double sided (the three layers of milky velumn were needed to create the copper trace/tracks etc). And yes, more sheets of velumn were needed for Solder Masks. Component Code drawings were also often done on the drafting table with its bank of 40w fluorescent tubes, for backlighting. By the way this is the reason why most CAD programs have you drawing on a light grey background, so as to emulate a 'backlit' light table. It's also why most of us 'oldies' turn the background to Black, to cast off those 'old ways' of eye strain and much soreness from a big job of layouts. Of course the same procedures were followed when a board was to be made using artwork produced using Bishop Graphics stickdowns. The big camera was used to make the exposing neg, either 1:1 or reduced 2:1 if the artwork had been done actual size. Then the neg was used for photoresist exposure and etching, etc. A simple 'contact' image reversal was often made and is still made for quickie NEGATIVES for UV Riston Exposures. Digitising for CNC drilling One procedure NOT fully explained regarding the onset of CNC drilling machines was the need for us board makers, to have to 'digitise' the X & Y positions of every one of those 'blooming' holes. This was done via an X-Y encoding table with rotary encoders on the lead screws. In the early stages the rotary encoders gave us 1024 bits to the inch of accuracy; later the 'quadrature' shaft encoders went to 1000 bpi and a small header was put into the PCB drill file to instruct the CNC drilling machine whether it should drill at 1024bpi or 1000bpi. Ah, those early fun days of going to Italy and shipping out the PLURITEC CNC drilling machines. The USA-made EXCELLON machines were much more expensive, and probably better (although this is debatable). All were mounted on 6" (150mm) to 8" thick granite slabs for rigidity, and all could have up to four (even six) drill heads on the X-Y travelling boom. Up until that point in time, all artworks had been Single Sided, made using the 1943 surface method -- 'masterpieces' in Black or Dark RED or Indian ink, or using the Bishop Graphics type stickdowns. In both cases black or dark red could be used, because the Ortho film saw Black or Red with much the same reflectivity. Hence the transition from Black crepe tape to Red precision cut tape a little later was of no major problem. Personally I tended to prefer the RED tapes, because I didn't have the creative skills of Terry Flowers and another kid appearing on the block, George Lizier to lay 'straight' lines etc One of the methods of laying out these now 'old fashioned' artworks was the fact that the RED or BLACK 'masterpiece' was laid out as per copper/solder side. This meant that 'text' could be direct applied to the layout using rub-on Letraset lettering. This was a simple method allowing various key items on the artwork to be identified as AC, DC, pin 1 and so on, done in simple 'place on and read' fashion. However all us 'oldies' were about to have our world turned upside down with the move towards Double Sided boards (with Plated Through Holes coming much later, and multilayer much, much later). So at this point we, artwork layer-outers had to re-think, and the transition took a year+ to easily get on top of. It took me about 2+ years to mentally adjust (not to smart). Anyhow, with the advent of the Top Layer of Copper tracks, commonsense dictated that this layer became the main viewable layer - just at it is on the finished circuit board (because all of the component labels and values etc are easy to read). So the BLUE top layer is then viewed correctly on the VDU, and the RED copper/solder side is now 'flipped' or in 'reverse'. To their credit PROTEL, did follow these rules and permitted us CAD designers to add 'strings' (text) to the board design them, then flip them backwards so when the RED copper/solder side was printed out to film it appeared with the text etc all now reading correctly. But I'm jumping ahead Precision Slit stickdown tape The next era of artwork design was the Red/Blue precision slit tape, again by Bishop Graphics. I think Mr Kodak was approached for his input to the eyeball spectrum and it appeared that Red and Blue were at opposite ends of the visible spectrum, as well as available simple wratten filters, (OA & #3) and so common films made of the day could handle them. RED was still used (or Black) as the colour for Single Sided boards, and as most boards to date were either '1943 early surface mount' or Single Sided with a few holes, the use of Red or Black donuts/tapes artwork required about the cheapest film, developer and time. So RED became the defacto standard for all Single Sided boards, even as the Bottom Solder Copper Solder Side Layer for Double Sided boards. The introduction of Blue for the Top layer was a commonsense next step and it became easy to lay down the Black pads for ICs and various other component pads for bottom (and top) to solder on, A single sheet of velumn then you would place RED tracks etc to become the Bottom Layer. Then you place your sheet of frosty velumn on it and laid your Blue tracks for the Top layer - plus any miscellaneous Top Layer pads, etc. BLACK pads would be COMMON to Both layers during photography (and a registration pin bar would be fitted just prior to photography, to make sure they registered accurately), while the RED and BLUE trackwork would separate to the two layer negatives because of the light colour and film stock used. Note that later the BLACK pads also came to be used for the plated-through holes, joining the top and bottom copper pads. (The term '43 Surface Method' was loosely given to those circuit boards which had the wire tails of components bent into a 'drip loop', rather like the 'skid' of a Bell Jetranger helicopter) and these are then put where the components are needed on the copper side of the circuit board. This is similar to what is nowadays regarded as a ' NEW ' method in some amateur radio magazines, called 'blob' board construction. It is obviously a rather slow method of assembling a circuit board, but as P'n'P (Pick and Place) machines were not invented yet, this was a way it could be was done in the low productivity era.) I was stunned to learn from a large US company (emails kept) that their design staff (who profess many years of collective experience) DID NOT KNOW these basic fundamentals of artwork preparation and colour separation. Anyhow... Orthochromatic Type 3 film will 'grab' the RED and BLACK layers, so next you left the Black pad layer on the backlit exposure frame and placed the BLUE pads and tracks (top) artwork layer over the Black pad layer. This time you had to use a much more expensive 'panchromatic' film to capture the Blue image, and it had to be handled in total darkness or a very expensive, hard to get KODAK #3 Dark Green Filter. Why is that? Well, the common Ortho Type 3 film has a very low ASA (exposure index) rating of probably 3 to 6, which is why we could work comfortably in a dim red lit darkroom. In contrast the Panchromatic film used for the Blue track/pad separations had an ASA of at least 32 (it may have even been 50-64-125) - so the penalty was that you had to handle the film from the frame to the developing trays in either total darkness or a very, very dim Dark Green light (15w bulb & KODAK #3 wratten filter) at the OTHER end of the darkroom. As a result the cost of producing the Negative for the BLUE Top layer was much higher than for the RED/Black Bottom layer. Not only was Panchromatic stock much more expensive than ortho stock, but the extra time needed in handling jacked up the cost as well. And extra time for the board maker inevitably resulted in a 'lightening' of the customer's wallet. It's true to say that the cost of Negative production for BLUE Bishop Graphics artworks was more than four times the cost of producing the negatives if RED / BLACK artworks were presented. So, defacto standards were established: RED was used for the much more common BOTTOM Layer of both Single Sided and Double-Sided boards, while BLUE was reserved for the less common TOP layer of Double Sided boards. As you can see these established standards for the colours of artwork design had nothing to do with the colours for human optical reasons etc, - they were a simple matter of the time and cost per item. But these became the defacto standards were later written up as the US275 standard. What this all means is that YOU can have your OWN standards, but please do NOT ask someone with years of experience in doing things the correct way, to view your masterpiece on the VDU if you have your own 'grandiose' colour scheme, if you do so, prepare to be rudely treated and dismissed. The experienced person who does follow the industry 'standards' cannot work out easily and quickly your Bottom (Solder Side) layer from the top (component side) layer (on your COMPUTER SETTINGS) - and they don't want to waste time and money working it out, go look for another job soon. COMPUTER AIDED DESIGN era Computer Aided Design or 'CAD' programs for printed circuit design were first introduced (I think) by Hewlett-Packard or Texas Instruments, and these first CAD programs were real 'dogs'. I saw the first one a long time ago and had a four-hour play with it. Sorry, but it really was a dog - slow, methodical and very procedure driven. It seemed to have NO resemblance to an electronic version of Bishop Graphics rub-on pads, tapes and puppets, so I went back to these for a few more years, until better CAD 'more friendly' programs came along. The NEXT real CAD program that arrived for me to use was smARTWORK, developed by a US company called WINTEK. This program had CGA graphics (not as good compared to today's VGA etc) BUT to my mind, it still has the fastest point-to-point routing algorithm for suggested track placement I have ever seen, even to this day. But it could NOT produce components - so it was a very slow, older way of pads and tracking placement. We put up with this for a year or two. AUTOCAD came out with a library extension for their famous engineering program, excellent it may have been, but cost wise, it was $$$ idiotic. Then (during 1984-86 I think) PROTEL arrived on the scene, invented right here in Australia. At first they appeared to be a tinpot company from Tasmania, but they seemed to be MORE aware of the requirements for better and far more accurate circuit board design, and even with their first package they succeeded very dramatically. Again the early versions were in CGA graphics, but they moved later very quickly into VGA. (640 x 480) The early programs Protel produced were PROTEL 1, 2 and 3, but the stalwart which really was the base standard for quite a while was Protel 3.03. This was the forerunner to AUTOTRAX 1.61, which to my biased (warped & luddite) point of view is still the best CAD circuit board layout program since sliced bread. By the way I don't get a brass razoo for my endorsement, but I do respect smarts and intelligence and bloody clever software. PROTEL also saw the need to 'amalgamate' the normal tech drawing schematics with the board artwork department for general book-keeping and records. This they achieved early in the scene with a program called SCHEMATIC EDIT A program which I have NOT used much at all. After all I design and make circuit boards, so I never got too involved with the pretty 'book keeping'. The feature I like about AUTOTRAX 1.61 ND is that although it's nominally a DOS program, it can be used in just about any operating system I have ever used: from MSDOS 2 - 6.2, even in the Windozzz series of operating systems, Win 3.1x, Win 95 (ugh) , Win 98se2 (excellent). I know a couple of board designers who still use it running it on Windozzz XP. I have yet to find a DOS emulator for Linux, but when I do I will try AUTOTRAX in LINUX some day. (can anbody help ?) PROTEL have moved on to further into the various CAD programs with extraordinary results. I am not totally familiar with these newer packages, but I've seen them in action. Yes, they are pretty incredible, but the last few programs seem to have drifted away from the initial CAD design arena to now be written in 'Nerd-CAD-glish' language. None of which tells me that those brilliant software designers seem to understand what a circuit board is really for, because of some of the non-tech names of the various steps in the actual design area. Also the VDU screen is really cluttered compared to that of Autotrax 1.61 - where I now have a full 21" monitor screen of my artwork design, not just the mini 1/3rd centre area crowded by all kinds of other info surrounding, including what is on Pay TV, I think. It is hoped that they re-name and re-do some of the steps in the process as I continually hear complaints from my customers asking for help in trying to produce a simple artwork. It would also help if they brought out instruction manual written in tech Australian like the old AUTOTRAX 1.61 book, and not this 'newspeak Nerd CAD-glish '. Anyhow with all my grumblings, I can still find no other CAD program to come anywhere near the PROTEL standards. However I must express my push to new kids on the block to look at Circuit Maker (including TRAXMAKER), as they are the closest Windows equivalent of the DOS-based Autotrax & Schematic Editor. Because of my ways and age, I am still a DOS luddite. I really detest the number of stupid mouse clicks needed to place a pad and track. Phew!!! Talk about regressive technology -- I am surprised that they do not suggest a lit candle, gas/kerosene lantern or Solar Powered room with pedal power. When PROTEL CAD became obvious to me, to be the 'defacto standard of CAD design' I gave away freely a few copies of their 'FREE DEMO DISC'. Well, they sooled their legal team onto me from Canberra, so for a few months bad PR was the order of the day. At RCS I did not accept any artworks drawn with Protel, legal or otherwise, as they suggested it could have been on one of my demo giveaway's. (which could NOT save) It amused me to have massive companies have to sit out the front when I was at Bexley, and they would have to re-tape up their artwork using Bishop Graphics over their PROTEL printouts, so I could make a four-hour prototype. Boy, they did grumble a lot about it -- and complained to PROTEL in Tasmania. Finally after two or three months Mr. N.M. of PROTEL worked out a way to resolve the impasse (after some of the big customers threatened to return their legitimate purchased software). I have no doubt there may be better useable, described out there to use, all based on the PROTEL CAD ideology, and possibly better written and easier to read/understand than 'Nerd CAD-glish'. But they all shoot themselves in the foot, because we (RCS Radio) do NOT have the legitimate ability to make circuit boards from the artwork they produce. Recent problems arising from an organisation in Adelaide made it clear that we are still NOT permitted to make circuit boards for school projects based on their maybe-better-explained CAD program but very, very poorly promoted. PROTEL (now called ALTIUM) decided to make their older 'baby brother of AUTOTRAX' DOS program EASYTRAX available as a freebie download off their web site. Then they made AUTOTRAX 1.61 ND available in the same way. To my mind this has destroyed much chance that any other non PROTEL CAD Software developer can get a foothold in the 'introduction to Printed Circuit CAD' market. AUTOTRAX is an excellent start for the chappies and chappettes to get going with. But if you are a new starter on the block, then perhaps 'Nerd CAD' maybe easier for you to understand. Not for me, I have NO doubt many designers start on the earlier freebie CAD software and then they move over (with much protest at the new descriptors) to the newer windozzz type stuff and produce some incredible designs. I have seen quite a few people make this transition -- and yes, I can convert Easytrax to Autotrax and vice-versa, as well as from Autotrax to Protel for Windozzz and back again. So I can make the circuit boards, and I do not have a need to better understand the latest 'enhancements' for changing, adding extra components etc. I have NEVER been impressed with PROTEL's bloody awful component libraries, which to my mind reveal their obvious lack in understanding the evolution in component design. But Protel did give us the ability to create our own library components -- so I have generated all these since 1986 based on a need for each job. The PROTEL software designer's nomination of pin 1 as the grab/move point for ICs is a 'Nerd CAD' position and shows they have never used Bishop Graphics puppets. I'm sure they simply do NOT understand the mirth (and violent hostility) when you go to 'move' or 'rotate' a component and it disappears/flicks off the VDU screen you are working on, out into the 'never never land'. Hence all my components have the grab point 'feducci' in the centre of the component. It is called 'common sense' -- a characteristic missing from 'Nerd CAD' people both in speach and text. Another minor annoyance with the newer Windozzzz packages is that you cannot put any miscellaneous 'hints' on the screen to remind you what you want to do in the next session. The Windozzz programs really spit the dummy and cannot handle anything apart from a genuine Protel component for application and use with tracks connected. A minor headache, but you learn to live with it and purchase a paper memo-pad. My mate found this out to his dismay when designing, using PROTEL's latest and greatest CAD drawing program, as he popped some misc comments onto the VDU as 'reminders' to double check the connections of a certain component. Well, his file did all sorts of strange things and was badly corrupted. It took him many hours using QDOS (an old hackers editor) to pick his way through each line and remove the offending 'self imposed screen' reminder tips, and bring it back to real CAD info only.. PROTEL's drill file One really excellent feature of PROTEL's packages is their ability to produce a drill file for our CNC drilling machines, and one that is 'edit- able'. This simple procedure does seem to have been forgotten in nearly all of the other CAD programs I have seen and not a simple ASCII file. Another nice feature with the Protel CNC drill file is the ease to get at and repair certain X,Y location drill sizes. A certain electronics magazine (which can remain nameless) had a particular small circuit board design with only a row of 10 LEDs plus an IDC header connector. One would expect such a board to have only 20 + 20 = 40 holes, but oh no! When the PROTEL program was queried about the number of holes in the board, it came up with a figure of over 1000 holes. Hmmm, I decided there must be a slight error there, so I opened up the main listing of the actual PCB FILE 4 source file, (using QDOS). There I found a LARGE number of Duplicate X-Y hole locations. All the duplicates had to be physically deleted, or it would have destroyed a 30,000 rpm air bearing drill motor. These high speed drilling motors do not like 'rattling around' in fresh air' drilling the same hole for say 100-plus insertions. smARTWORK may still have the fastest point-to-point track router algorithm I have ever personally used and seen, and it was (and still is) a class act. But every pad, track, via and fill had to be 'hand' placed and it was tedious, with no ability to invent or place 'components' so as to do a typical IC block dump. You also had to invent a component code overlay, another time labour job, unlike being automatically 'free' by PROTEL. Another disadvantage with smARTWORK was the manufacturing of the Circuit Boards, in that the board manufacturer, must have a paid-for package (like to days EAGLE & WIZARD CAD programs) on their computer system for the artwork file to be actually 'read' and maybe 'correct', then played out to a plotter etc so as to go to the next dept. and then have a Negative made. It also did NOT have a drill file that could be used with CNC drilling machines for mass production. To get around this the customer would drop off his copy of the program (with key files) , all before email, and pick up the floppy disc with the circuit boards, this did slow board designs. When PROTEL arrived they not only had a DRILL File (which was easy ASCII editable) but also an incredible command using two alpha characters - the 'LA' or 'Library Add' command. What a feature. This really is the command of all commands, because it enables you to create your own 'components'. But a warning here: 'Will Robinson, do not exceed 200 components in any library' - or 'poooooof' the library will explode, disappear, and be gone forever into the ether. So learn well the other swear word: 'BACKUP' and have your component libraries 'stashed' elsewhere -- in case of disaster. The playout from either a 600dpi HP LaserJet 4 printer (is there any other brand?) or from a 6000dpi Linotronic typesetter achieves an output onto OHTP (overhead transparency) (or direct to DIAZO) film quite excellent for the next stage: namely the black artwork positive to Ortho Type 3 negative process, in the darkroom. The output Laser/Linotronic device produces Black Carbon on clear images but the general use of normal inkjet/bubblejet printers is a sure sign of limited mental adjustment -- because, how can a vegetable colour (black) withstand a UV converted film (or direct on)? Yeah sure, you can print it out twice and sandwich the OHTP's together to get some sort of black - but in general, this still gives a very poor 'contrast ratio'. From memory, only one inkjet printer, an Epson 1290, I think can print really Black pads and tracks, or a fake negative of any practical use. The rest you can use, but you end up with a lot Narrower 'time window' for exposure to Riston laminate in the later steps of the procedure. - - - Stop PRESS, I am experimenting with a couple of ink jet printers using Pigmented Inks and DYE sublimation, reason being, ORTHO film is becoming nearly impossible to obtain, so in a few months , I will have a new announcement regarding 'certain' 'branded' inkjet printers and ink. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - About the component libraries on the CADROM CD With the Component Libraries you have here on this CADROM CD I strongly suggest you load them into your computer, into a temporary holding directory and then you start by copying the 'nothing' Lib file called RCSBLANK.lib elsewhere, renaming it as you wish (for example BLOGS1.lib). Then extract some of the more popular used components into your own named library, and you may even need two libraries for local stuff. I have made a library of the most common ones (I use) and it is called RCScompX.lib. The components in this library are taken from the various libraries RCSTRAXa.lib and up to RCSTRAXz.lib. It is not difficult to change libraries when using the program, even when your 'masterpiece' is up on the screen, meaning RCScompY.lib and RCScompZ.lib are also quick libs. I strongly suggest that you DO NOT ADD to any of the RCSTRAX.libs in case some one near and dear to you decides to upload the libs off the CD back into your well created new components, and you have spent a zillion hours of various items deleted. I know this has happened to a few of my mates in recently in Melbourne, England and the USA (twice). My libraries may never be fully completed. CADROM2 went from versions 2a to 2f before I realised that I had an inherent problem in the component libraries. I had drawn them with 'five-thou yellow outlines' which is quite excellent for home designers and makers of personal circuit boards. It's also quite OK for us here at RCS Radio, because we use a NEW silk screen every time that a customer wants a (white) Component Code and (green) solder mask. But it's not so good for larger commercial board makers because they use photo-imageable UV cured inks. Yes, it's still a blanket of white screened on and dried, but the image is a piece of film, and the area in question is UV exposed, developed, 'fixed', washed and dried, then any extra processes are carried out. As said earlier, we use a NEW screen every time, as it is fairly easy for us to cut out the old screen then re-lay new mesh, tighten, glue, trim cut, ready to go again. But with the bigger commercial players the 'screen' is only a carrier of the dots, so they wash out the screen with a Gernie blaster (well about 98% or better) and re-use the screen many many times. But re-used screens cannot easily screen the 'dots' of a five-thou image, with the result the white component outlines do seem to have missing 'bits' Now I personally went over to one of my larger (much larger) competitors and ran some tests. They can sometimes screen six-thou about 97% of the time and if they use seven-thou outlines, then every component with these outlines come out nearly perfect, But With EIGHT-THOU the results were 100% perfect every time - even on sloppy (lost tension) silk screens the results were the same as with my five-thou outlines on a 'new' screen. So I am accepting as a defacto standard 'eight-thou' outlines for all my components. That meant 'Library Exploding' every one of my 300+ components, then globally changing all the 5-thou outlines to 8- thou, then block define, library add ('LA') and bingo I had re-done all my components. Phew, what a task! By the way, 10 thou or bigger is quite ok, (if FIXED as component standard) but when you use mixed outline sizes 10, 12, 15 etc, that are LARGE then your eyes tend to be blinded by the glare. You have poor visibility to see even your RED (Bottom Solder side) or BLUE (Top copper trace) layers. Also when you use 10, 15 or 20 thou variant outlines, they DO look VERY SLOPPY - -like Brown's cows in a paddock with no uniformity or visual equality. The bottom line is: stick to a 'standard', either US275 or your own. So the task, the CAD overview so far, has been done. However I will be always adding to the libraries, and you, the CADROM owner will be able to send it back every 6 or 12 months for a very small price to get the latest updates as a CD Disc exchange. ONLY Master Discs will be service honoured. Also don't forget if you want a 'special' component, please make email contact with me, so I can draw it up and add. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Hints on making 'Better PC Board Artworks' 1. For most of the pads for resistors, capacitors etc, you should think of a base standard diameter of 75 to 85 thou. If you have the room, then look at up to 100 thou, as larger pads do tend to adhere to the circuit board better when doing 'repairs' with 36 or 40 as a default 'standard' 2. Many new kids on the block when designing a large blank area of Copper design a 'fill' with the WRONG method: using a thin (less than 20 thou) track/trace and basically doing a 'basket weave' to fill in the area where the fill is wanted. They seem to use a silly narrow overlap between tracks, which means umpteen tracks and results in a massively large file size. This method I always liken to using a 1" (25mm) paint brush to paint the side of a country barn. In short, I simply call it 'dumb' or maybe 'dumber'. A recent a/w with basket weave 'fill' was some 200k, but when 'modified' into a 'drop and plop' full fill, it narrowed to the file size to 30k. The RIGHT way is to Place a Solid Fill. In Autotrax 1.61 this is so easy: just 'P' for place, 'F' for fill, then use the mouse cursor to define the Two diagonal corners of the fill area, and 'plop' a full fill is done. I am sometimes asked about CAD software to purchase, and they specify it must have an 'autorouter', so I raise my eyebrows and know 'another ho-hum person thinking he/she can emulate in five seconds what was/has been worked on for the last 60 years' -- novel and cute. Well this is an unpaid advert for PROTEL's CIRCUIT MAKER , which as explained earlier is really only a combination of the earlier DOS Autotrax and Schematic Editor, now in 'Windozzz'. So you still have to draw up the schematic, double checking the point to point connections, then make a 'net-list', then feed that into the TRAXMAKER section of CIRCUIT MAKER, and away it goes depending on prior 'rules' regarding component Placement/Spacing and size. It then produces an artwork that can be 'adjusted' or vastly improved when using DOS Autotrax, as the files are transferable and easily QDOS editable. Whilst there are others now in the marketplace - EAGLE, WIZARD, and a host more -- the only problem is that we as a commercial board maker are NOT permitted to use them. Not only because we are NOT licenced to do so, but we do not have either a legal or shonky copy for use. Many so called free versions on the web are NOT Compatible with the purchased versions. One of the cleverest tricks that has come to mind is a company in the USA which gives away free CAD software, poverty pack style of a PROTEL program, but insists that you get the boards made in the USA by them as there are NO drivers for download for local manufacture ! It is an excellent con, and I admire it. Normal designers out here in Australia, Industrial/Hobby have a great giggle at that, as a 'no brainer' for local manufacturing, and I must agree it suits backyarders, who have unlimited time & money to play with. ------------------------------------------------------------- Summary of Circuit Board materials, COPPER CLAD (Coated) Laminates It appears the very first material used to make printed circuits was 1/4" (6.5mm) bakelite sheet, with a milled-out groove where the tracks were to go. The panel was placed on a 'cake rack' in a tray of water, and molten copper was poured into the milled-out groove. When it cooled, the various components were then soldered onto it, with cometimes another sandwich of bakelite, was placed, screwed, bolted onto the finished 'now Top'. Next came the 1/8" (3.2mm) bakelite, (Bear in mind that at the time components were big and heavy, so a heavy base substrate was needed.) The TRIPLEX had a thin sheet of copper foil 'hot press glued' to one side. The copper foil was designated '1 ounce per square foot', defining its thickness in terms of weight per unit area. It was also described as '35 grams per metre square'. The first substrate (the insulating material supporting the copper foil) used to be known as XXX. (Triplex) Then the insurance companies got in on the act and wanted/created a Flame Retardant classification -- basically how long it could sustain a fire after the 'heat source' was removed. This Flame Retardant classification was abbreviated to the 'FR' rating. Bakelite would have been classed about an FR-0 (if it had been classed), as it can really 'supports' combustion. Anyhow FR-0, FR-1 and FR2 products evolved, with anti-flame 'stuff' in the 'fillers' with the most commonly used materials being FR-3 and now-a-days FR-4, which is the defacto base standard for all boards. So if FR-4 is the base material used to make all boards nowadays. It is mainly supplied in a thickness of 1/16" (1.5mm) and it comes to board makers in standard sheets measuring 4 x 3 feet (1220 x 915mm). Hence the reason that nearly all modern CNC drilling machines have 18" x 12" drilling beds - corresponding to a standard 4'x 3' sheet cut into eight even sized plates. There is material rated at FR-5 and FR-6, made for producing boards specifically for use in 'very sensitive' RF equipment working above 1GHz. These boards are usually Teflon/Fibreglass composition or pure Teflon (PTFE), but always mega-bucks in cost. Whilst the most common board material thickness used these days is 1/16" (1.5mm) the invention of 1/32" (0.8mm) was introduced for aircraft lightweight use as well as 1/64" (0.4mm). Even thinner 0.2mm and 0.1mm material has also been introduced for use in making multi-layer boards. MULTILAYER boards A Multilayer board starts out as Double Sided blank, then drilled, with quite a few very thin and pre-drilled Single Sided boards hot glued and bonded together as a 'sandwich' using many tons of pressure. This board is then oven dried and the glue in the holes is dissolved. (Sometimes they may have to be re-drilled to grab the 'vias'. Final plating of copper plus chemical interlinks of copper for PTH features, and solder follows and the board then goes through finals of solder masking, component code screening and so on. It is quite a lengthy process. Multilayer boards were first needed because of the limited number of better 'discrete' or 'LSI' components. But as technology evolved and LSI became common, then the need for many layers to carry the data etc, was removed because many of the connections were made inside the LSI comps. themselves. The common ones these days are usually three or four layer, but in the beginning I was involved in making six and eight layer boards. I believe Italy used to produce up to 16-layer circuit boards ! The individual layers must have been very thin. (Think about it, 16 layers of 0.1mm with glue and copper tracks making the final board approx 2mm thick -- a tad fatter than the common 1.5mm or 1/16".) , what a nightmare to fault find. Double Sided board material usually has '1oz per sq foot' copper foil on both sides of the FR-4 laminate, but when Plated-Through-Hole circuit boards are made, most board makers use '1/2oz' copper because during the plate-thru process an extra 1/4oz to 1/2oz is plated ON during 2nd stage plating straight after either electroless copper or carbon. Board makers who follow 'standards' use 1oz copper on most boards, but if not specified in your order you may end up with 1/2oz copper. This results in a lessening of the current-carrying capacity of the tracks. For the current carrying capacity of tracks in variety of copper thickness, refer to my 'AMPTRACK.exe' program. Some TV's being made in Asia, I have seen must be using 1/3oz or 1/4oz copper tracks. Just you try to repair or resolder those tracks ! A friend of mine at Bexley had to repair the local banks promo tv set, and when he complained about the ability to solder to, I used my solder thickness 'ping' meter, and it did NOT even register as 1/2oz, fact. So in summary, what is available to us board makers are the following: 1/2oz, 1oz and 2oz copper weight (thicknesses), or 17um, 35um and 70um. If needed, any extra copper can be plated on, but at a greater expense to (you) the consumer and not easilly verifiable, a simple matter of trust. ------------------------------------------------------------------------- RCS Radio - Robert J Barnes - sales@rcsradio.com.au January 25 2007