Dictaphone - Express Talk Digital Dictation System

Electronics

Engineering Career (page 1)

Links To Electronics - Engineering Pages:

 

1  2  3

Express Talk prototypes today, on display in my lab

Express Talk literature collection

Express Talk recorder interior.

Looking for a job in late 1979 after graduating from FIT, what I really wanted was to design consumer electronics. Well, by this time most of that had already gone overseas with some work still being done in California. I ended up working at Dictaphone in Melbourne FL (or later the companies that would buy Dictaphone or the facility I worked in) for almost 29 years. I always knew it wouldn’t last.

 

Unwilling to move across the country, I saw Dictaphone as similar to consumer electronics, as they did make things that real people would pick up and use. With no experience and just out of college, I accepted a job as Test Engineer. I really didn’t enjoy that type of work, so I began offering redesigns of existing sub-assemblies that either reduced cost, assembly time or offered greater functionality. Even though it was a manufacturing facility, I began to reshape my job into design engineering.

Dictation machines at that time were based on a continuous tape loop that would gather in layers at the bottom of a bin. They nicknamed the machines “tanks”. A traditional reel-to-reel machine won't work because you have to allow a new dictation to be recorded at the same time a transcriber is playing back a previous dictation from any random place on the tape. While the “tanks” worked, it was mechanically complex, allowed a maximum of one dictator and one transcriber at a time and required maintenance such as head cleaning.

At the time, hard disk drives were just starting to be affordable. They were now up to 10MB! I worked some numbers and found this was enough space to record a reasonable amount of compressed audio with a bandwidth sufficient for speech. I began experimenting on my own time, working on various compression techniques. I new there would be lots of data moving around and did not even bother with my favorite processor, the 6502. I jumped right to the MC68000 which was quite a powerhouse at the time. It ran at 8MHz and could directly address 16MB of memory! It was hard to imagine that you could ever need or afford that much memory. This was the same µP being designed into the first Apple MacIntosh.

While I was working on this, Dictaphone did end up sourcing a small hard disk drive based dictation system from JVC. It was not expandable, but more importantly it still used our old analog dictate and transcribe stations. What I was working on would be expandable allowing up to 6 people to use the system at a time in any mix of dictation and transcription. But what really set it apart was that the A/D and D/A conversion was done in the stations and everything was sent digitally on a RS485 bus. Today, that seems obvious as everything is digital. But in the middle 80’s, it was a big step. This is what made my design the world’s first all-digital dictation system. During it’s product life I designed many additional peripheral devices such as telephone interfaces, handheld and handsfree dictate stations, etc. Also, I was always making provisions for ever increasing disk capacity. See .pdfs of the sales brochure and the in-house newsletter where they talk about my design.

As you can see from the internal photo of the main unit, the recorder, the motherboard had the 68000. It also had 6 slots where communications boards could be plugged in. Each of these bds. had a 6502 µP, RAM, EPROM & and RS485 transceiver. This would create one 125Kbaud bus and would translate into one user, either dictate or transcribe. I used the fact that the 6502 really only needed access to memory during half it’s bus cycle to create dual ported RAM using only standard SRAM. The 68000, then, could sneak in during the unused part of the bus cycle and access each plug-in board’s RAM. This access was therefore transparent to both processors and allowed audio and control data to flow through buffer and handshake flag constructs I set up in RAM.

 

I had to develop the main 68000 software and at the same time write the software for the 6502s on the cards. It didn’t stop there, though, as I also had to write 6511 (an early Rockwell µC based on the 6502) software for the Dictate Station & Transcribe Station as they all interacted with the protocol I was putting together as I went along. All this on the prototype hardware for all these units which I designed somewhat in parallel. Looking back, I’m glad I did this in my 20s, as I don’t think I could keep it all straight today.