CADHouse Diaries #1

Posted by bob on June 8, 2022

Back in the 1980s, Benjamin (let’s call him Ben) had started a small business that designed microcomputer boards for industrial control and monitoring applications. One of his first needs was to find a company that could reliably translate his schematic diagrams into Printed Circuit Board (PCB) artwork. He would then send that artwork to a different outside company that fabricated those bare PCBs. A third outside company would stuff and solder the full PCB assembly and then send them back to Ben for testing.

After a few cycles, Ben came to greatly value the relationship with his PCB design company, which he generally thought about as his “CADHouse,” where “CAD” stood for Computer Aided Design. Such CAD systems were highly efficient as compared to manual black tape and Mylar PCB layouts. The CADHouse PCB layouts were created on large and expensive mini-computer based systems. The CADHouse had a likewise expensive service contract for maintenance and repair of their minicomputer systems. Unfortunately, the traveling repairmen for that service contract often did not carry spares for all of the important pieces and seemed to be relatively poor at troubleshooting the underlying problems.

During a series of phone conversations between Ben and the CADHouse, the CADHouse manager suddenly had a brilliant idea. “Hey Ben, you seem to understand these computer systems pretty well, and we are getting such poor help from our service contract, could you come look at this one problem? Coincidentally, if you can fix the problem, it would help us get your current design done sooner, since our primary CAD system is broken.” Even though it meant a 90 minute drive each way, Ben felt like the trip might be fun and if nothing else, he would get to look behind the curtains a bit, learn about the challenges his vendor faced, and maybe build some good will with the CADHouse team.

The first problem Ben went to fix was a large failing CRT (Cathode Ray Tube) monitor that the CADHouse used as their primary interface to see the layout as it progressed. Something like a television set, this monitor converted a high resolution raster image from a display controller into glowing phosphors on a screen. The screen would turn on and display the expected image, but at random times, the image simply faded slowly to a completely dark screen. Sometimes the image would come back after a power cycle of the monitor, but not always. Likewise, the working/not-working times seemed to be quite random.

When Ben arrived, he brought in his portable oscilloscope. He expected that he might find a problem with the horizontal deflection circuit of the CRT, since that system was used to create the extremely high voltage needed to accelerate electrons from the neck of the CRT over to the screen phosphors. Ben removed the rear metal cover of the monitor, exposing the CRT and circuitry. Simply placing the scope probe inside the monitor chassis showed him a waveform that met his expectation for a horizontal flyback signal. Indeed, the monitor was in its working state.

But as he sat behind the monitor thinking about the general design of TV sets, the screen brightness suddenly began to fade away to a completely black screen. Yet the horizontal signal was still there and unchanged. Also, unlike some deflection system problems, the display did not seem to distort its shape as it faded.

Ben repeated the experiment again, this time ignoring the oscilloscope. He just looked inside the monitor and tried to absorb any hints or clues. He asked the CADHouse engineer to turn off the room lights before turning the monitor ON again. This time, when the monitor display faded away, Ben finally saw something change. When the monitor was first turned ON, Ben could see a faint red glow at the end of the CRT neck. This glow was the hot filament. (This filament provided electrons which could accelerate across the CRT and then illuminate the screen by slamming into phosphors coating the inside of the screen.) But when the display faded to dark, he could see the filament glow slowing dimming out.

Ben expected to find an AC transformer that converted 60Hz line energy into the low voltage needed by that CRT filament. By tracing the wires to the CRT neck, he eventually found a transformer, but not the one he expected. The filament was wired to a PCB that then connected to a heavy insulated wire that went up and wrapped around the outside of the high voltage horizontal flyback transformer. The hundreds of volts of horizontal deflection signal was stepped up to perhaps 35,000 volts for the CRT anode, but was also stepped down to about 6 volts for the CRT filament. The circuit seemed so simple. The flyback transformer was clearly still operating, yet the filament current was not getting through to the filament.

Ben powered off the monitor and studied where the wires attached at the single-sided PCB. These were fairly heavy gauge stranded wires. He could see a faint ring of discoloration between the PCB pad and the wires. Now he was quite sure of the diagnosis.

Once he could see it, he could fix it.

He pulled out a soldering iron and commenced to reflowing all of the filament connections on that PCB. He took his time to make sure that fresh solder flowed cleanly between the PCB pad and into the stranded wire. He added some tie-wraps to make sure the wires had a short and secure path to the PCB.

Retesting the monitor showed that it now stayed working with no intermittent fade-outs. Over the next few weeks, the CADHouse confirmed that the monitor was now completely stable.

Many years later, Ben shared this tale with an expert at a TV set manufacturing company. That expert smiled, and shared that (from painfully similar experiences) his company had learned to use a crimped eyelet to solder and attach stranded wires to single-sided PCBs. The eyelet provided additional mechanical strength for heavy connections—beyond the native strength of the solder joint.

This simple repair provided the start to a long and productive relationship between Ben and the CADHouse.