A step-by-step guide to hardware manufacturing

By Tony Diaz

Have you ever wondered what goes into the making of a peripheral card for your Apple II, or any other computer for that matter? From beginning to end, many things take place. This article covers each phase so you can get a basic idea of what goes on.

There are a lot of considerations when making hardware, the two most important being the price to make each board, and the ease of manufacturing it. Included in these considerations are the components that will ultimately be used in the finished product. An early tidbit… when Woz was working on the Apple II, it was decided to use Dynamic RAM, even though it was more expensive, because it was easier to design for and required less support components, meaning ultimately, less money would have to be spent. Besides, like everything else, it was hoped that DRAM would decrease in price as it became more popular, not to mention the volume discounts Apple Computer would be able to count on.

Every idea starts in someone's head and gets transferred to paper (or some other medium) where it is looked over, leading to the prototype phase. In the case of hardware, one will usually draw a schematic of the logic, a kind of theory of operation.

The main component of any hardware is its "printed circuit board", commonly called the PCB. The PCB is just a bunch of wires ironed on the surface of a fiberglass sheet and cut to size. The wires are referred to as traces. A "breadboarded" version may be assembled, which is a generic PCB that is usually covered with holes and some common traces already run for things like power and ground.

This is an example of a wire-wrapped board. It's the back of a prototype Ninth Wave RAM card which became the Applied Engineering RAMWorks memory card for the Apple IIe. Yes, those are all wires! Imagine trying to remember where they all go.

This is an example of a wire-wrapped board. It's the back of a prototype Ninth Wave RAM card which became the Applied Engineering RAMWorks memory card for the Apple IIe. Yes, those are all wires! Imagine trying to remember where they all go.

Working with a breadboard is also known as wire-wrapping. This phase is where you have oversized thick sockets with long leads. The legs of a component are called leads, and are often composed of wire of size 30 gauge. An example of 30 gauge wire would be the metal part of a tie-wrap from a loaf of bread or package of buns. About one inch of insulation is usually stripped from the 30 gauge wire and the end of the wire is put down the hollow end of a tool that is then placed over the protruding lead. You then rotate the tool and the wire will wrap around the post (lead)—hence the term "wirewrapping"—causing it to be very tight and enabling a good connection.

Conceivably, you could pay for and receive product in wire-wrap form, but it is uneconomical to make products this way. The failure rate would be greater, the assembly time would be prohibitive, and it would drive the cost through the roof for most people.

The breadboard process is usually a time of discovery as well—you find that changes in the circuit are needed, or things drawn in and implemented are unnecessary. This is another reason to make a breadboard version of something, especially if it is a complex design or something you have not worked with before.

A tidbit along these lines—electronic circuits can have a personality, and the Apple II is not immune. Case in point: the Basis 108, an Apple II clone from Germany. It was designed pretty much with only the schematic from the Apple II Reference manual and from that respect it was 100 percent compatible. In the real world, it turned out that it could be called "too compatible"—sometimes making a change at point A has implications not at point A or B but at point G.

Once you think you have the bugs worked out and you're ready to make a PCB for your new product, you have to take the final revision of your schematic and turn it into something that (for an Apple II) is no bigger than 2.5 inches by 8 or so inches. The final layout is likely to be a double-sided PCB, meaning that it has traces on the front (component) and back (solder) sides only. The Apple Disk ][ Controller and Apple IIe motherboard are examples of a double sided PCB. You could also do a multi-layer PCB which would have four layers or more.

The Apple IIGS motherboard and Focus PCB are four-layer PCBs. A four-layer PCB is most commonly used to separate the power and ground onto their own layers. This division is usually done to increase reliability on sensitive circuits. By having power and ground available nearly anywhere within the design, filtering is provided, as well as the need to run less traces on the PCB. On nearly all PCBs, the traces on one side run along one axis and the other side runs along the perpendicular axis.

This is an expansion board that's a little farther along in the development process than the prototype RAMWorks, the never-completed SoundMeister Pro. If you look closely you can see examples of tight wire-wrapping.

This is an expansion board that's a little farther along in the development process than the prototype RAMWorks, the never-completed SoundMeister Pro. If you look closely you can see examples of tight wire-wrapping.

Now consider this: if you recall, one of the most important considerations was the final cost to produce the product. Let's start by gauging the cost at 10 cents per hole as part of your PCB production cost. A 6502 has 40 pins; that's 40 holes that need to be drilled and plated on the PCB, meaning $4 to place that 6502 on the card. The SCSI ribbon connector has 26 pins: $2.60 (you start to get the idea).

For the purposes of this article, I will focus on a Super Serial Card PCB that I have here. It is different from the released product in that it has a 26-pin header for the cable that connects to the serial port or DB-25 connector on the back of the computer. The released version shipped with only a 10-pin header, 16 pins fewer than mine, or $1.60 less of holes in the PCB, using my 10 cents a hole rule of thumb [See Picture 1]. Back in the early days of the Apple II, it was rumored that Steve Wozniak scrapped the whole design of the Disk ][ analog card (the PCB inside the Disk ][) and started over because he was sure he could get it done with fewer holes—the net savings was one hole after a few days of work.

When looking at a finished product, you will see some holes out in the middle of nowhere on the PCB (holes with nothing placed in them). These holes are called VIAs, as in "the trace travels via…". The components themselves also require holes. The PCB layout designer is tasked with using as many of the component holes as VIAs as possible instead of needing extra holes on the PCB. The fewer the VIAs, the more room you have, the more tightly you can place everything, and the less the PCB will cost.

The PCB cost is composed of the number of holes drilled, the sizes of those holes, the area of the PCB, and the amount of gold plating to be done. Gold plating is usually limited to the slot-edge connector and costs 10 cents a "finger" for this example, a "finger" being a pair of the slot edge pins, front and back. The Apple II has 50 pins in each slot, or 25 pairs. That means the gold plating costs $2.50 in the price of the finished PCB.

Looking at the same SSC PCB, there are five pairs of fingers that are not present. There is nothing electrically attached to these pins so there is no need to provide a contact point. This reduction saves 50 cents of the potential cost. Over a large run of PCBs this can add up significantly.

The hole sizes are another factor. PCB manufacturers are typically set up for three drill sizes per job. This SSC PCB has two hole sizes. The VIAs and component holes are the same size and there are larger holes with no connection to them at all (these are usually tooling holes). The machines that prepare the PCB for soldering have to hold onto something, so these holes are used for assembly.

PCBs may go through one revision from breadboard to final production release, or they may undergo several. You want to be sure you're done, because making a full production run of PCBs and then finding you have to make changes on each one after you receive the finished lot is no fun. Every ROM 01 Apple IIGS board has "by hand" modifications done to it because it was cheaper to do so as part of the assembly and production than it was to go back and work these changes in—the problems were discovered too late in the design phase to fix.

Apple Computer is probably going to get a better deal than Applied Engineering might because Apple will make more cards per batch and more cards over all. All of this is a factor in the PCB process. With the first version of the Sirius RAM IIGS, there were four different PCBs done in the prototype phase before the card was declared final. The SoundMeister IIGS card is a good example of a card with a high VIA and component hole count.

I learned quite a bit about the VIA elimination process from the time I had 1,100 holes on an Apple IIe memory and RGB combo card to the 640 or so holes on the Sirius RAM IIGS. The 1,100 holes on the IIe card were reduced to below 400 several years later with a revamp using SIMMs and some programmable logic parts.

The Apple IIe was an improved Apple II along the same lines: from a component count of over 100 parts down to under 40 (not including discrete components such as resistors and capacitors). The board size was also reduced by about 1/8th and functionality improved. Part of the component reduction in the IIe is attributed to the design of custom circuits to do the job of, say, 10 others. Like the DRAM used on the original Apple designs, money saved here could be spent elsewhere for an overall more reliable product—the less things installed, the less things to go wrong.

What was, what might have been: At top you see the everyday Super Serial Card that is very commonly in an Apple IIe. Below that, a never-completed alternative Super Serial Card.

What was, what might have been: At top you see the everyday Super Serial Card that is very commonly in an Apple IIe. Below that, a never-completed alternative Super Serial Card.

This streamlining can be referred to as modernizing the design for manufacturing efficiency. As technology evolves, methods change. Building through-hole PCBs was becoming more costly because more companies were going to surface mount. For example, the Apple IIGS was modernized to be the first Apple computer to make extensive use of surface mount technology. The ROM 3 was also a change in this direction—the 256K x 4 DRAMs used on the ROM 1 were more expensive than the 1MB x 8 DRAMs used on the ROM 3.

The Apple IIGS platform needed 1MB as the base memory, as 256K was not enough anymore. By eliminating the need to build separate RAM cards and include them in the systems (as well as costly "by hand" modifications being done to each motherboard), the ROM 3 replaced the ROM 01 similarly to the way the IIe replaced the ][ Plus.

There you have a little insight about what transpires to get hardware onto your desk. Many, many projects never make it through half of this process; many more turn out totally different from what was started.

Some get shipped and have to be recalled or scrapped because not enough was done during the design phase. Until next time…


Apple II hardware guru Tony Diaz lives and works in Oceanside, California. To view a Web site related to Tony's hardware Tech-torial, Juiced.GS readers should point their favorite browser to: http://www.apple2.org/HardwareProduction/