On the previous page, we saw the Big Trak schematic and circuit board. Now, we'll take a closer look at the Big Trak’s chips.
The Big Trak’s brain is a 4-bit TMS1000 microcontroller running at approximately 0.2 MHz (at 6 clock cycles per instruction that’s 0.033 MIPS). Members of that chip family were used a lot in calculators and consumer appliances. It cost less than $3 at the time, in quantity.
The program is burned into the 1024 byte ROM and is not modifiable. Because the program is in the microcontroller itself, not a separate ROM chip, it is not possible to replace the microcontroller with just any another TMS 1000 chip. You’d need to pull one from another Big Trak.
Texas Instruments TMS1000NLL 4-bit microcontroller from 36th week of 1979 (7936) and 5th week of 1980 (8005) with MP3301A (Big Trak) program. Note in the example of the right, the part labeling drops TMS1000 and adds 4955, which was the model part number for the Big Trak.
I don’t know if there were different versions of the Big Trak program. The “MP” portion of the label on the chip refers to the particular program. So far, I’ve only seen MP3301A as the program label for the Big Trak.
The bottom number on the chip label is the two digit year and two digit week of manufacture of the chip. For example, 7936 is the 36th week on 1979. The label establishes only that this Big Trak can be no older than this date. Since custom chips are produced in large order batches, the toy manufacturer would have several months supply all with a similar date code (depending on how many chips could be made in a week). Thus, there could be a significant gap between the date on the chip and the actual assembly date of the Big Trak.
The TMS1000 chip contains a whopping 64 bytes of 4-bit RAM (which equals 64 nibbles or 256 bits or 32 bytes of modern 8-bit RAM). Not 64 gigabytes, not 64 megabytes, not even 64 kilobytes -- just plain 64 bytes!
Some websites mistakenly claim the TMS1000 had 32 bytes of 4-bit memory or even as little as 64 bits (not bytes) of memory. But, that can’t be the case. There are at least eight commands (FIRE, FORWARD, SPIN LEFT, HOLD, SPIN RIGHT, BACKWARD, REPEAT, OUT) and all but one require a value between 1 and 99. Storing eight commands requires at least 3 bits and a number between 0 and 99 needs almost 7 bits. Therefore, it would take at least (3 command bits + 7 number bits) * up to 16 steps = 160 bits of memory for the longest user program. I tried a bunch of random combinations of commands and numbers to make sure pattern compression wasn’t a factor. So, the chip does indeed have at least 160 bits of RAM.
I was struck by the low capacitance values of the capacitors. If you add up all the capacitors in the Big Trak, there’s less than 1 microfarad of total capacitance. Compare that with all of today’s robots running around with 2200 µF capacitors across their battery inputs. Also, the Big Trak doesn’t include any fancy low-ESR tantalum, metalized polyester, or aluminum organic capacitors either!
Part of the reason the Big Trak can get away with this is the separate power supplies. While four 'D' cells power the majority of the robot, an independent 9 V battery powers the TMS1000 brain. This electrical isolation helps prevent electrical noise from the motor, speaker, and lamp from resetting or interfering with the microcontroller.
An old axial capacitor that looks like a resistor.
The lowest value capacitors on the Big Trak (C18, C20) are in axial packages with color bands that look like resistors. That’s something you don’t see anymore.
The Big Trak’s brawn primarily consists of a 75494 Hex Digit Common Cathode LED driver. This chip is also known as SN27423, SN27914, or National Semiconductor equivalent DS75494. Usually, this chip would drive the individual LEDs that form numbers on a numeric LED display, like on a pre-LCD calculator.
The term “Hex” means this chip has six inputs and outputs. The term “Common-Cathode” means its outputs were designed to sink (0 V) the cathode end of the LED. This chip is not capable of outputting a high (greater than ground) voltage.
Think of the 75494 chip as a six-pack of transistors -- although it actually is slightly fancier as the inputs can be controlled by the weak output pins of early microcontrollers and logic devices. And that’s the reason why it exists inside the Big Trak. It takes a weak output signal from the TMS1000 and delivers heavier current (several hundred mA) to thirsty electronic consumers such as the lamp, speaker, and Big Trak Transport trailer motor.
For ease of upgrading, I decided to desolder the TMS1000NLL and SN75494N chips and to reinstall them in DIP sockets instead. Desoldering isn’t easy, especially for components with more than a couple of pins. Destructive desoldering (cutting off the component pins with a Dremel and then desoldering pins individually) is easy, but here we have irreplaceable components that we don’t want to destroy in the process.
Removing DIP chips soldered directly to the board by heating the board with a heat gun while pulling the chip away with a spring clamp.
My TMS1000 chip pulled out cleanly and neatly. I was so proud!
However, the SN75494 chip had a slight accident, as I must have lightly squeezed and released the spring clamp as I was pulling the chip out. The edge of the chip broke off as the clamp tightened again on the corner, sending the chip flying to the back of the work bench. But, that wasn’t the worst insult that the SN75494 would have to endure.
I brought all of the parts upstairs after extraction. Along the way, the SN75494 must have fallen to the tile floor. Back at my desk, I couldn’t find the chip and so I figured I had left it downstairs by mistake. As I entered the tile hallway, I stepped on the chip!! That’s right. 25 year old chip. Irreplaceable. I stepped on it.
And, you know what? It made a crunching sound when I stepped on it.
16-pin DIP with bent pins damaged from being stepped on and a broken edge (see arrow) from a spring clamp during desoldering. The pins have not yet been cleaned of residual solder and the exposed portions of each pin show dark oxidation from age. On the right, the pins have been carefully straightened and the old solder has been smoothed or removed.
After I got done with the obscenities, I removed the chip from the sole of my shoe (yes, it was embedded).
I then began the long process of cleaning; starting with gently bending each pin back into position. Then, I cleaned the chip with an ultrasonic cleaner with detergent, smoothed/removed the remaining solder with a soldering iron and no-clean flux, cleaned it again with an ultrasonic cleaner and isopropyl alcohol, and then finally nudged it into a DIP socket (which also helped align the pins). Believe it or not, the chip works beautifully.
Now let’s see how the transistors drove the Big Trak’s motors...