Early PCB CAD systems prove to be an uphill battle, but things eventually work out.
This is the third article in a series about my career experiences in the PCB industry. The second article focused on the start of my design service bureau.
When my company, Computer Circuits, moved from manual to computer design, we unexpectedly lost all our customers. The companies who relied on manual design simply did not want to change – they didn't trust it, and they could not make changes to the artwork themselves once the initial design was done. Plus, my rates more than doubled. The good news was that quite a few larger companies had switched to the Redac Mini and we got some of their overflow work.
The computer system was a DEC PDP-11/34 with 64 kB of RAM and 5 MB RL01 disk drives. It is hard to believe that a design could be done with so little resources. In fact, larger designs had to be split up. Otherwise, they would overflow the memory. The maximum memory that could be used on an upgraded PDP-11/34 was 256 kB. Today, a Windows PC can have 128 GB of memory – 500,000 times more than the PDP-11/34.
The Redac Mini had a VT11 vector graphics display, and everything was represented by green vectors or dots. It also used a light pen. Today, some people get carpal tunnel syndrome from using the mouse too much. Imagine holding a thick pen in the air, touching the screen at the desired location, pressing a button near the top of the pen to select, holding it down, moving the pen across the screen to edit, then releasing the button – over and over, for eight to 10 hours a day. We eventually found that it was easier to tape the button down! Moving the pen close to the screen (about a half-inch away), it would select. Then we’d move the pen to do the edit, and pull the pen away from the screen to release the object. While this was still tiring, it was an improvement.
Before the end of 1981, we upgraded to Redac Maxi systems, which had larger disk drives and more memory, as well as a color raster display by Megatek with a tablet and stylus instead of the light pen (Figure 1). We also bought a Gerber 6241 photoplotter (Figure 2). To support the additional equipment and employees, we moved to a larger office space.
Figure 1 Megatek monitor with tablet and stylus
Figure 2 Gerber 6241 photoplotter
Most of the designs needed to be data-prepped from the schematic. This was a task that involved manually creating component and net lists which got punched onto a coded paper-tape using a teletype machine. It would then be read into the PDP-11/34 to create the digital database. With the upgrade to the Maxi system, we could type the netlist directly into a file.
After the design was completed, we generated the Gerber 274-D PCB layout data. The plotter flashed a light through an aperture in a wheel containing various size apertures for the pads, and smaller ones for traces. The plotter was huge and noisy. We were located on the second floor and it shook the room as it jerked back and forth while exposing the film. These plots would be sent to the PCB fabricator who would use them to create the copper etch masks.
One day, I met the owner of the realty business which operated on the floor below the plotter. Shortly after our friendly introduction, I was pointedly informed that the noise and vibration were completely unacceptable. Being a good neighbor (and not interested in getting kicked out of the building), I changed our process to plot only at night. Oh, the joys of running a small business.
Occasionally, a customer would have a deadline that required special attention: Every customer wants the design ASAP, but sometimes, the need was real and extreme – the success of the company or the job of the engineer was at risk.
In one instance, our decision was to work two shifts, which was actually not uncommon. Once the data-prep was done, the daytime designer started the placement and finished it in one shift. The night-shift designer started the routing and worked on it all night. Coming in early the next day, the daytime designer was supposed to continue the routing. When I checked the progress, I saw that almost none of the routing was done. The designer said, “The routing wasn't done as I thought it should be, so I deleted most of it and I am now routing it the way I imagined.”
My initial reaction was anger because of the deadline we were up against, but on closer inspection, I couldn't disagree completely with the decision. When a designer does the placement, imagining how the routing will be done guides component location. Of course, the night-shift designer didn't know what was in the head of the first designer. Using two designers on the same layout failed. This familiar lesson impacted my perspective on future software development.
As much as I enjoyed managing Computer Circuits and working with the team, I became, year by year, ever more intrigued by the design software itself, and yearned to be involved in its development.
- An Excellent History of CAD
- The Gerber behind the Gerber file format
- A lifetime designing PCBs: Early design adventures
- A lifetime designing PCBs: Job shopper to service bureau
- Next blog: A lifetime designing PCBs: From design to software
— Charles Pfeil is a Senior Product Manager at Altium, working on definition of their products with a primary focus on routing tools.