TMDS -The First Television Monitor Display System
by George Michael
At Livermore, we had been considering various schemes to provide graphics to all our computer users at as low a cost as possible. Going from a digital picture description to a video representation would allow use of low-cost TV sets for presentation. Ordinary display consoles for each user were far too expensive, to say nothing about the incredible computer burden in refreshing many displays. We had considered Storage Tubes instead of CRTs, but the costs were far too high, and the images were too static.
The obvious solution was, somehow, to use TV sets as the display organ, and some form of low-cost memories to service the refresh requirement. At that time, low cost was best embodied in some form of disk storage.
Our first version of the TMDS used a disk that held 64 tracks on a Ni-Co coated platter with a contact head per track. This was built for us by Armin Miller at the Data Disk Company. It was very inexpensive to the point that we could plan on having many disks. We were unsuccessful at trying to store a complete video picture per track, and with a nod to reliability, ended up using four tracks for a complete video picture. Even so, that gave us an initial capacity of 16 video channels per disk.
The TMDS went through four upgrades in the course of its approximately 20-year lifetime. The final version had a capacity of 256 channels, of which about 200 were populated. We used RAM for the last two versions because their bit-for-bit cost had dropped below that for disk. The output of the video channels was fed into a crossbar switch that expanded the number of channels by a factor of ten. Thus, we had the potential for some 2,000 video channels of which any 200 could be in use. Each user's office was supplied with a television set and a keyboard (usually an ASR 33 Teletype)connected to the central computer system. The TV set was initially connected via high-grade co-axial cable. Later, connections started to employ optical cable. Ultimately, the costs per terminal ranged between $1,400 and $4,000; a factor of about ten less than the cost of an equivalent graphics system. As TMDS matured, individual users added all sorts of TV-compatible devices, such as tapes where motion sequences could be collected, printers, and stunt boxes that could mix the output of several video channels to produce either gray scale or color images.
The ordinary home television set is called a 525-line system, with the odd-numbered lines painted in Frame 0, and the even-numbered lines painted in Frame 1. However, only about 489 lines are used for the video display. The unused lines are not displayed and represent the time needed for the operation of the flyback oscillator to reposition the painting beam at the start of each Frame. In the case of the TMDS, we used TV sets with the capability of 558 lines, of which 512 were used for display. In theory, the best displays come when the distance between two adjacent points on a line is the same as the distance between two adjacent lines. When this Interlace Ratio is one, a displayed circle will look like a perfect circle. Another characteristic worth noticing is that the refresh cycle of a TV is independent of the displayed content; clearly not the case for an ordinary graphics system--a graphics display cycle is "content-sensitive" while the television display is not.
Any user will tell you that good hardware is important, but the real measure of greatness is good software. The TMDS, one of the three best developments during the first 25 years of the Lab's computer activities, was effective because superior text manipulation software was available. These utilities were named TRIX and TRIX-AC. TRIX employed features found in both SNOBOL and LISP. It also included graphics functions to process display lists, perform vector-to-raster conversions, etc.
TRIX-AC implemented an unusually complete set of macros to support pattern matching and text editing commands. These routines were written by Henry Moll and Alexander Cecil, and, literally, thousands of people used them daily. They were designed for use with any computer terminal, but were beautifully matched to the speed of the TMDS for viewing and altering both text and images.
For more information on the TMDS system, see Mel Gregonis' interview.
For technical documentation of TMDS systems, see the following scanned manuals:
- TMDS Meeting
- TMDS Manual
- TMDS II User Manual
- Showtime
- Status Display Screen Shots
- Block Diagram