Z-80 Space-Time Productions Single Board Computer

My Personal Experiences at Tandy Computer Assembly

I graduated from TSTI/Waco (now TSTC) in Waco, Texas in August of 1983.

After taking a short break I started job hunting in San Antonio. My start into electronics technology was with Tandy Corporation at a plant called Tandy Computer Assembly which was located at 1515 Goliad Road, San Antonio, Texas. Although none of the components of the TRS-80 computers were ever fabricated there, all the Model III computers and Model 4 computers as well as Model I and IIB modems were put together there. Also, an early form of Plug-N-Power modules were assembled there as well. In the upstairs store-room were boxes and boxes of Model III BASIC manuals and boxes and boxes of silver and gray Plug-N-Power cases that we used for various small projects.

When I arrived for work at the plant, the Model III production had been shut completely down although there were several of these still connected to test equipment and all the "Function Testers" for the Model 4 were created from Model III circuit boards and case bottoms which were sandwiched together. At one point in time I located a box of Model III cpu boards that had been sent back to the "Tech Area" for repair that were caught in transition when the III shut down. I was told they were "old junk" but to hang on to them in case a tester went down somewhere.

TRS-80 Model III

The Model 4 production line was well under way. I worked there for just a little over one year. We received all the components for the Model 4 from a plant called "Tandy Advanced Products" in Fort Worth, we put the units together and ran them thru testing. They were then boxed up and sent back to TAP where some of them were pulled for QA testing.

TRS-80 Model 4

This was how the plant functioned:
Board Test
PCB boards first arriving at the plant were initially tested. These boards came in cube shaped cardboard boxes with cardboard dividers between each board. There was two huge wooden racks where the PCB's were first powered up and tested. Boards were loaded into these racks and connected to power supplies. A video connector was passed from board to board and the output verified. As I remember the boards were left in this rack for some period of time before being passed onto the Main Assembly Line. The video screen would say "Cass?" when first powered with no FDC connected. Any failures here went straight into boxes of failed boards which were then carried back to the Tech Area.
This Board Test area was run almost entirely by Vietnamese women who happily chattered their language all day long. They were gorgeous women who wore fire-engine red lipstick and spoke broken English. One day when the sonic welder in the bezel area was running, bats came down from the suspended ceiling. The women from Board Test put empty pcb board boxes on their heads, screaming all the while the Floor Techs chased the bats out.

Main Assembly Line
A long snake-line assembly line with rollers was where the plastic bases of the Model 4 were started. The AC power cord and plate, power supply was mounted. The diskette drive box came next and the PCB's were mounted inside two aluminum case halves for FCC RF shielding. This completed part was installed at the back of the plastic base standing up as though it were an 8-1/2 by 11 sheet of paper laying in "Landscape" mode. These aluminum covers were very sharp on the corners and could cut you pretty easily. Cable harnesses were put in and screwed down to the base and connectors hooked onto the boards. FDC's were also installed and diskette drives into the bays, where ribbon cables were attached.

The CRT's were installed into the top half of each case and it was carefully laid on its side on foam to avoid scratching. A scratch on the plastic case meant doom from the QA people, and that computer would be sent back for replacement.
One thing I remember fondly about this part of the process is that on rare occasion, a CRT would be dropped from the assembly line where it would loudly implode with the sound of crashing glass. The entire plant would burst out in loud applause. I never figured out why this happened, but it made everyone laugh.

Often you would see people in the upstairs observing offices looking thru the windows right after this. Oh yeah, that was another thing. There were several small rooms in the upstairs both front and back halves of the plant where managers could peer thru 'almost one-way' glass and observe the plants workers. They'd make note of the slackers, of course. If they forgot to turn off the room's light, you could see their faces thru the long horizontal slotted windows. Big Brother was watching you.

The video boards were installed into the upper halves of the housing and cable harness connected down to the power supply. Then the unit was passed on down the line. The keyboards were then installed in the bottom front part of the base plate mounted by 4 screws on the corners. Then the flex cable was hooked up to the PCB, making it a 'live computer'.

Disk Test
There was a place along the assembly line where workers ran tests on the Model 4's diskette drives. There was a series of read/write tests done by in-house code on test diskettes and target diskettes. Occasionally you had the computer which would get dubbed "Disk Blower" when it would damage or over-write the source test diskette. They'd swap the drives out and try again. If it failed again, the FDC was switched, and if again, the main PCB would get pulled and sent back to the Tech Area.

One thing to note here: We discovered an interesting bug in the Model 4 computer. There was some spacing between the machines on the assembly line, but sometimes at Disk Test they'd be bunched together for more thru-put. It was here the Floor Techs had one of their most interesting problems. Model 4's that were sat side-by-side would often have multiple diskette errors. They swapped diskettes which failed from wear at times, tried several things. It took a couple of weeks to finally make the correlation to what was happening, but it was this: When two Model 4 computers are sat immediately side-by-side, the video board inside the left uppper housing of the right computer was very close to the sensitive diskette drives of the computer to its left. This interference from the high voltage flyback of the right computer's video board was getting into the left computer's diskette drives, causing the errors. There was some experimenting with more shielding, but I don't recall ever having an absolute "fix" for this problem. The video board already had a metallic mylar shield which was tied to a green grounding wire. The diskette drive bay was an aluminum housing which shielded the drives themselves (but not the ribbon cables).

Burn In
This was a huge area of the west end of the plant with serpentine assembly line rollers where the assembled Model 4's were lined up. I don't remember how many but I would figure there was probably between 100-200 computers burned in at a time. People loaded the Burn In program from diskettes onto the computers. This was an odd piece of code that I sincerely wish I had source code or copy of. The memory was set by PAL so that it was one complete 64K bank of ram. The program wrote $00 thru $FF repeatedly thru the whole memory map. It would go thru and verify the contents of each location from bottom to top, then it would increment each value in ram. This memory area included the video ram, so each display had the character set containing all the alphabet, numbers and special characters and would advance about once every 4 or 5 seconds. This gave the appearance of the screen scrolling the letters. The top line of the computer contained the time the program had been running in hours:minutes:seconds and some other odd information, I can't remember it may have just said "Tandy Computer Assembly Burn In Test" or something like that.

Sometimes units would fail at Burn In, their screen would be blank except for a graphic box which which surrounded a message in the center stating what memory location was written, the data written and what was read back. This would be recorded on the cardstock with that board and it would accompany the PCB back to the Tech Area.

Hi Pot Test
This was a single operator test station at the end of the assembly line. A really cute girl with long blond hair worked there, I think her name was Michelle something. She would hook the Model 4's coming out of Burn In onto a tester. This device tied the Hot and Neutral wires together and applied a massive voltage, I think over 1KV between there and the protective earth Ground pin to make sure the computer wouldn't leak over a certain amount of current. Once it passed this, the computer was ready to be boxed up for shipment. Computers passing her test were put on long gurney type carts for transport to the next area.

Shipping
In this east end of the plant area, the Model 4's were again inspected for any scratches in the case. Plastic bags were fitted around them, and they were put into cardboard boxes with styrofoam corner protectors. These boxes had the Tandy logo all over them and a picture of the computer. As near as I can recall, these boxes were white with light green and light blue print and graphics on them with some black for the part numbers.

Receiving
Also back with Shipping was an area where all the parts coming in for computers. They were inventoried and stored on shelves for delivery out to the Main Assembly Line. Parts were inspected for damage and flaws, cosmetics and some function here.

Floor Technicians
On the north west corner of the plant was an area where about 5 or 6 of us worked. I worked here for a period of time that I can't recall. We made sure that all the testers in the facility, power supplies and other devices around were running and in good order. We often came in on days when the plant was closed to build new testers. We had access to a great Craftsman table saw and tons and tons of plexiglass from 1/8" thru 3/4" thickness that we used to build sides of cases for mounting monitors for testers and in some cases we built boxes from plexiglass to hold power supplies and other circuit boards down on test "pogo" pins which would press into the solder-thru's for appplying signals or reading voltages. We had some kind of chemical there, I can't recall what it was, but you could wipe some on the edge of a piece of plexiglass that had been cut, and set it on another piece of plexiglass and it would weld it and the weld would be as transparent as the rest of the plexi. It was fascinating stuff, and we had to use gloves to handle it. I have a few chemical names in mind, but not sure which one it exactly was.

The Floor Techs were the best bunch of guys I worked with the whole time I was there. I can't remember too many of them but Malcolm Robertson comes to mind along with a guy named Freddie and another one named David. I hate dead brain cells. Anyway, the Floor Techs were all friends, we used to park backed in at the "Grassy Knoll" {see diagram} open the doors, hatches and play music while we relaxed in lawn chairs and ate our lunches, threw a little Frisbee afterwards. This was a great break from the eating area, where about 25 people in the plant did not smoke - ha. I'm exaggerating! But it was a nuisance. So we enjoyed the outdoor lunch breaks together.

Tech Area
The very back of the plant was a long rectangular room with workbenches along the north and east sides of the room. Here technicians received the boxes of failed CPU PCB's from the front of the plant. Strangely, this was the one area of the plant walled off from the rest of the production facility. At each station, there were two blue monochrome screens on either side. These were from Model III computers, and had the keyboard parts sawed off of them. In the middle sat a 20 MHz oscilloscope. Right in front of you sat a Function Tester which was made up of two base plates of Model III's that had been screwed together with all the sides open. Inside was a modified Model III board. It contained the rom (how I wish I had this!) which ran the menu and tests for the Function Tester. A 40-pin ribbon cable came out of the back of this tester {you'd plug it into the CPU socket of the Model 4 board under test}, and a keyboard was mounted in the front of the bottom plastic base. You had too hook this ribbon cable up to the keyboard input of the Model 4 pcb under test. The top base plate, which was upside down, was covered in a cut out piece of anti-static material. The rest of the work surface was also covered in antistatic as well. The display on the right was another cut-down Model III and you hooked the video from your Model 4 under test onto it to see what was happening. A power supply mounted in the upturned base powered both the Model III Function Tester board and the UUT.

I have to say a lot of praise about this area, because this is where I started my work as an Electronics Technician. It took me a bit to start really learning to troubleshoot electronics, but if you worked there long enough, you'd really get some amazing troubleshooting skills under your belt. We were provided with a production copy of the schematics [I still have my copy to this day]. I worked on the series of Model 4 that was "pre-Gate Array" and left the plant before I saw the first of them come along. These were being called "Gatorade" boards even before we saw them.

We worked some interesting problems back there, which really honed my understanding of TTL logic gates, and gave me a good idea for short circuits caused by the lead trimmer blade bending socket tips over onto traces or nearby pins. One problem we looked at quite a bit was labeled "Jail Bars. This was so named because of vertical beat frequency visible bars on the video displays. We were replacing a number of cheap dial-capacitors located at the top of the PCB for this problem, but there were times other components were involved and simply replacing that cap didn't fix them. We had a couple of boards like that which were 'Challenge Boards' which the guys would pass around after they'd caught up on their quotas and fiddle with.

Funny Tales from the Tech Area:
One thing I laugh to remember about this area is during Christmas, we were for the first time allowed to hear music. There was no radio allowed. Carlos had a cassette recorder used to CASS load boards that were in repair for tape loading or saving failures; and someone had brought in a 90 or 120 minute tape of carols. We were still listening to Carlos' Christmas tape well into the middle of January when the supervisor finally made him turn it off. And then the silence was back.
Ram jumpering was a common failure of the Model IV boards, often they'd come outfitted as 16K ram computers, but the jumpers would be set for 64K. The bad boy was when it was a 64K computer, and the jumpers were configured for 16K. The 16K x 1 bit dynamic rams used +12 Volts and -5 Volts, the 64K x 1 bit rams used only 5 volts. If +12 and -5 was applied to the 64K rams, they'd get hotter than fire. I distinctly remember the day a guy burned his forearm laying it across the board with the rams mis-configured. He probably still has that row of tiny rectangles scarred into his arm with the part number visible.
In another incident, there were a couple of Op-Amps on the board which demodulated the cassette tape for program and data loads. I remember one guy hooked up this board, then left the area for a restroom break. While he was gone, the board - having its op-amp installed backwards - caught fire. We all smelled it, then noticed a tiny flame and a thread of smoke headed straight up. We were all chastised to unplug any repair boards from power if we were leaving the bench after that.

There was also a machine in here that had been built to help troubleshoot Model 4 boards. It consisted of a table with a huge vacuum mounted tray of pogo pins for holding the UUT, and on the front had a tape drive. The keyboard was an aluminum box with the keyboard and a 1 line LED 14 segment display, and a small thermal printer for controlling the tester. It was a 6502 based machine and used a tape drive on the front right side which was constantly running during the test. I was priveledged to work on this machine for a while, but it was a flop. I had to cut the crystal oscillator signal trace, remove the Z80 and MC6845 before it could be tested, and all this had to be put back on exit. The unit was loud and would vacuum hold the board to the pins (which I think was also breaking traces all the while). The tape would spin for a while. It would prompt me to put its logic probe on specific chip points before moving on such as "PROBE U13.3". Then it would spin the tape for a while and finally spit out something like "U5.7 SA0" which meant U5 pin 7 was stuck at zero. Or it would give a list of pins that were supposedly 'Stuck' in some logic state. I'd put a wire on the crystal oscillator, put the main chips back on and send the board over to a tech who would usually tell me it was something totally different wrong with the board. After a few weeks of this, they put the machine to rest and moved me out to the Floor Techs, where I would find happiness after all.

Engineering
I worked here for the last couple of months of my stay at TCA. The department was headed by a great Z80 assembly programmer named Dale Deviney. He was listed in the IEEE membership during the 80's but I haven't been able to locate him at all. He was really a good guiding resource for me in coding Z80 and worked with me on some interesting projects. James Porteous was also in this department working as an engineer, he was kind of the dirty-grunge more typical of the 80's computer nerd mental picture we get. He was a nice guy, and knew his code well. We also had two software guys in there I think were from Poland; they seldom spoke English and spent their days coding away on Model 12's using Xenix and who knows what. They were very much white shirt and tie pro's in a t-shirt and jeans environment. One of them was named Januzs something, I can't remember. He was involved in development of a power supply tester that I built, he coded it with some of the most convoluted BASIC I've ever seen in my life. They were assisted hardware wise by an African-American named Charles - He and I had both 'graduated' from the Tech Area. He was always very well dressed and professional. I have a feeling he's gone on to some much higher profile work.

PAL and ROM Programming
There was a long rectangular room, which I suppose had been the butcher area of the former grocery store, which had large glass windows. The few people inside wore white jackets, hats and gloves. In this room, there were gangs of Data I/O brand programmers. In these areas, staff would burn the PALs and program ROMs which went on the Model 4 boards. They were driven by Model III computers (one wonders what they used throughout the plant prior to this, if it 'twernt the Model I) and the files loaded from 5-1/4" diskettes.

Modem Assembly
The TRS-80 Modems models IB and II were assembled at TCA as well. This was an isolated area off to the east of the main part of the plant. Ladies plopped components into bare pcb boards, they were checked for polarity and part number, then sent in a line across a wave solder machine located just to the east of the Tech Area. As they boards came off, they were assembled into the case halves and the label overlay put on. These modems were cute little buggers for 300 baud, but I remembered the day they took some of us into a room and announced we'd be building a 1200 baud modem. Obviously out of the technology loop, we were surprised that speed of tone could be carried over standard phone lines which were limited to 3000 Hz on the high frequency end. This would be the DC-2212. You'll note here that the Modems I, IIB, and DC-2212 all come in the same case with different black stick-on overlays. The switch cutouts were on all models, but the various PCB's didn't have every switch. The black plastic overlays covered the unused switch positions. They used those nifty little bat handle toggles with the shiny silver barrels. One technician in the Tech Area named Charles Bloom repaired the faulty modems. He had a couple of phone lines at his bench for testing modem to modem. One time I remember one of the modems was mis-dialing the number. In the tinny speaker, we could hear an elderly woman say "Hello?" which was followed by the buzz-saw sound of the modem trying to connect, her still saying "Hello? Hello?" in the din of racket.

Ultrasonic Welder
There was a station where the black plastic inner bezel of the top half of the Model 4 was welded into the white upper half of the case. I don't remember this device running all the time for some reason, maybe only certain batches came not already welded. But what I do remember is being able to hear the high pitched squealing when it was in operation, and the bats that lived in the suspended ceiling of the plant hated it. On days it ran, the bats would figure out a way to come below the ceiling and they'd drive all the women crazy. They screamed and panic'd, it was usually a fun day when they were going to run the Ultrasonic Welder.