原创 How it used to be: CAD, CAE, and Netlisting

Editor's Note: This "How it used to be" piece was emailed to me by a reader named Glen. I'm still trying to persuade him to send me a pair of "Then" and "Now" photos showing him as a young engineer and as the seasoned professional of today.

In the mid '80s my employer was using CAD in the drafting department for PCB layout, but the engineers still drew their schematics with a drafting table, D-size mylar, plastic symbol stencil, HB pencil, and – of course – a pencil rebooter (aka an electric pencil sharpener). The component delete function was performed with an electric rotary eraser.

The move to CAE was filled with many highly-frustrating learning curve errors. Workstations that crashed and ate the unsaved schematic when one was too fast on the mouse (the 'fix' was to shut off the cursor to slow down the engineer). Components with hidden power pins that were supposed to automatically connect to power and didn't (or did, but they connected to the wrong voltage). Thousands of text warnings and errors to wade through to find one that was meaningful. Schematic sets that took three hours to print (but could at least be photocopied easily). Compiles that had to run on the same workstation that the schematic was drawn on.

I could go on, but suffice it to say the early days of CAE were not easy, and all of these "features" caused a lot of trouble to engineers who were more concerned with the actual electronic design than fighting with cranky system issues. Of course, management thought CAE was a wonderful time saver.

However, some things that early CAE did do well after all the compiling bugs were eventually fixed were the back annotation (re-naming reference designators to match physical PCB location), creating the BOM (electronic parts only, mechanical parts were still added by hand), and netlist generation. Especially the netlist generation. Since the drafting folks used CAD, there HAD to be a netlist – they did not look at the schematics anymore.

This is how we created the netlist before CAE:

First we asked the drafting department to make a set of blueprints of our D-size mylar film drawings. Then we commandeered a meeting room for several days and hung the paper schematics on the walls all around the meeting room. A bunch of engineers would walk around with yellow highlighters and trace out every single net on every sheet, identified by name. One person would sit at a table and write down (yes, by hand) all component and pin reference numbers as the engineers called them out. At the end of several days the final check was to make sure every connection on the schematics had been yellowed, and hope the transcriber had not made any errors. Checking this would have taken almost as long as the original activity.

Then after the drafting department had finished the layout we had to perform back annotation. Manufacturing insisted that reference designations had to be in a sequential order on the PCB instead of randomly assigned while drawing the schematics. The drafting department would return a hand-written from/to list of all the new component reference designations which we then had to manually copy to the original schematic mylars. In one case, a colleague went to fetch the list and the draftsperson fished the only existing list out of his wastebasket ("What the heck do you need this for?").

The back annotation included any gate swaps that the draftsperson did to facilitate trace routing. Usually was not a problem since it normally did not matter which gate in a package was used. But this is not true for packages in which banks of buffers are tristated through control pins. Fortunately we caught this while transferring the gate swaps to the schematic. ("But you told me I could swap gates!")

After final layout we had drafting plot a set of Gerber films for inspection. Ever tried viewing black and clear Gerber plots through each other on a light table? Engineers did not have access to the CAD equipment to view multi-multiple layers in separate colours.

Yes, early CAE had its good points as well as bad points. A humourous incident after we went to CAE was that one of the management people told the office admins to stop ordering 0.5 mm pencil leads since the engineers were now using CAE. We quickly got that situation rectified.