0.8 Taking the speak n spell into production  (Page 4/5)

On top of all of these issues, we had a bit of a bug in our algorithm that caused a speech quality issue. We had designed the unit to take on additional vocabularies. This was in response to the focus group’s input that the vocabulary size wasn’t large enough. It would also allow us to give students more words, generating an additional revenue stream for the product. Unfortunately, if you pressed the key to enable the alternative spelling words in the module, and a module wasn’t there, the Speak N Spell would start talking with whatever data it was getting from free air. We called this “garbled speech”. Sometimes it would finish and come back and sometimes it would just keep going, on and on until the batteries were removed and the unit reset. We fixed it as quickly as possible but it caused a lot of heartache for all of us before it was over.

As an example, we received a letter and a cassette tape from a Speak N Spell owner on the subject of garbled speech. His letter simply said “my Speak N Spell going off every once in a while and starts giving random sounds, but, then it says a dirty word. I am sending you a recording of it so you will know what it says. And, no, I will not return it to you.”

Let me go back to the bug the night guard found and give a bit of explanation. We knew that the devices would need to be burned in prior to shipping. So, I had the PCB laid out so it could be powered through the module port. The module port is where a user would normally plug in additional vocabulary ROMs. But, for burn in, we could plug in a modified module which would power up the unit and defeat the automatic power down feature. What we found to be the issue is that we had used a signal diode rather than a power diode on each of the modules. But the production line rejected that solution and chose to use a different solution. They decided to use spring clamps to hold the "on" key down all of the time rather than replacing the diode. I chose to let them use their solution rather than continuing to complain about mine. This made life much easier.

If you don't know much about Texas geography, let me explain where various events were happening on the Speak N Spell program. It will help you appreciate the effort to take it into production. Here is a quick summary:

• Dallas. This is where the research and market communications team were. This is also the headquarters for Texas Instruments. Everyone knows where Dallas is, right?
• Lubbock. This is where the development team lived along with the plastics shop that manufactured the case, besels, keyboards, etc. Lubbock is in the panhandle of Texas and the home of Texas Tech University.
• Houston. This is where the IC's were designed and manufactured. I now live in Houston - if anyone cares.
• Midland. This is where the Speak N Spells were initially manufactured. It is about 150km south of Lubbock. It is named based on the fact that it is exactly halfway between El Paso and Dallas.

We were spread out all over Texas. The only good news is that if this program had been done today, we would have been spread out over the globe rather than just Texas. But it did make troubleshooting bugs and fixing them difficult. I'll give a couple of examples in the next paragraph or so.