0.4 The bugs  (Page 3/4)

“I know that you would like me to return the Speak N Spell so that you can verify what it is saying and fix it for me. But, I want to keep it as it is. I’ve sent you the cassette so that you know what it is saying.”

That seems to have been the only foul mouthed Speak N Spell that we shipped.

As soon as we determined the problem, our software development team, Glen Thornton, released a new version with the handshake included. That was the end of the software version with the garbled speech bug.

Garbled speech (hardware)

A second version of the garbled speech phenomena ended up being a hardware bug. It exhibited itself as randomly powering up in a mode that generated garbled speech. This is one of those times when I had to manipulate the various design teams. I sat down with the software engineers on the program and said to them “I know that this is a hardware problem, but the hardware guys are not willing to admit it. Could I get you to help me by finding a fix for the bug? And, please don’t tell the hardware guys you are doing it as it will make them mad.” I then sat down with the hardware team and said “I know that this is a software problem, but the software guys are not willing to admit it. Could I get you to help me by finding a fix for the bug? And, please don’t tell the software guys you are doing it as it will make them mad.” So I had each working on a solution independently. I then worked for a production solution independently with the product engineers on the production line.

I found that for some reason, if I put ceramic capacitors on the communications lines between the controller chip and the synthesizer chip the problem would go away. As I spoke to both sides (hardware and software) neither could give me a good technical explanation of why the capacitors solved the problem. But, as it did and was an easy fix on the production line, I introduced it to the production flow, first as a repair technique. Later it was made a permanent addition to the production line as it was cheaper (once we modified the PC board to allow auto insertion of the capacitors) than making it a repair procedure. As I had done many times in the past, the ceramic capacitors I called out were ones with a large number in "excess" stock. Because they were in excess stock (we had more than needed to keep the production lines up and running) the effective cost of the capacitors was zero and we didn't have to wait for a new shipment of parts to fix the bug.

As we got further into the production run of the Speak N Spell, we found the line shutting down periodically for this reason. At one point when the line went down we finally realized that the wafer fabs had switched recipes with another product. It was easy to correlate the recipe change to both products having issues in production. That particular time I was the featured presenter at a meeting with our senior management. The agenda was simple: the production manager presented how bad the design was, then the quality manager presented how bad the quality was and then I got up to defend the design with the expected result of my being fired. Fortunately I survived and we moved forward with a solution to the problem.