I thought I'd throw some more old pictures of my ancient projects at you. Click most pics to enlargenate them.

Crustcrawler Hex-a-pod robot:

This is my Crustcrawler Hex-a-pod robot. I purchased the kit sometime around 2003, just before I got into the hobby of collecting and restoring arcade machines. I was going to add Wi-Fi and video/audio telepresence, a GPS sensor and other sensors and then program it to be controlled remotely via a PC web-app or browser or a Wi-fi enabled remote controller device. Once I started to really get into it, I realized that it would be a ton of work that could literally take me a few years to complete. I didn't think that I would get enough personal gratification out of the project so I shelved it and was soon working almost exclusively on the arcade stuff.

The Crustcrawler company was absorbed into the Parallax Inc. company and the robot kits are still sold at the Parallax webstore. The robot below has 3 servos per leg so that the legs have a 'knee' function. This allows 3 degrees of freedom rather than the usual 2 degrees of freedom. This robot can be programmed to walk sideways as well as to virtually eliminate most or all of the foot-scrubbing that 2-DOF hex-a-pods usually suffer from. Programming walking gates that take proper advantage of the knee servos is challenging to say the least but the resulting increase in walking efficiency is impressive.

The 18 leg-servos are all of the double-ball-bearing type and are surprisingly strong. I didn't like the original fulcrum design for the up/down motion leg joint. It was a simple machine screw and the leg panels just had simple holes that the screw passed through meaning that the legs had to pivet and ride directly on the threads of the screw. This produced a lot of friction and unpredictable loading as well as increased wear. In turn, this would reduce the payload capacity and increase servo power consumption. I replaced the simple screws with a hardened steel shaft that also included ball bearings, ball thrust bearings, and collars and precision spacers. The resulting assembly is obviously more expensive but it is immensely more efficient, reliable, and robust.

The Pan/Tilt kit installed at the front end of the robot started out as the standard CrustCrawler kit but I then modified it to allow it to rotate 360 degrees. To do this, I added the two gears (ratio 2:1), a new base plate to mount the servo and the output shaft, a new bushing with internal ball bearings to carry the output shaft and a thrust ball bearing to carry the weight of the Pan/Tilt platform. The purpose of this was to be able to mount a miniature camera and some sensors to the P/T front plate and to be able to look up and down and all around the entire robot. Obviously this would be of great advantage when remotely driving the robot.

The control system that is presently mounted is the original system that I bought with the robot kit just to get started. Granted these boards were considered advanced in 2003 but there are far more advanced boards available these days for the same money. Check out Parallax Inc. as well as Pololu. The boards shown include a simple "board of education" that has a Stamp2 module, a 2x8 character LCD, 4 pushbuttons, and a pot. The other two boards are serial servo controllers. The system runs programs written in Parallax Stamp2 Basic, the editor/compiler/downloader is a free download from the Parallax site.


Ferrite Bead Non-Volatile Memory 1972:

This old circuit board is a 4 kilobit non-volatile memory module from an industrial computer that was built in 1972. The top board provides addressing logic and driver transistors for the memory board that is underneath the top board.

Before we get too far, let's put the memory capacity of the board into perspective with today's devices. This board stores 4096 bits of data. One kilobyte of data is 1024 bytes which is 1024 x 8 bits or 8,192 bits. So, two of the cards shown below provides 1 kB of storage. That said, it would therefore require 2 million of these boards to provide 1 MB of storage. A gigabyte is a thousand times a megabyte and so that means that 2 billion of these boards would be needed to provide 1 GB of storage. Considering that many thumbdrives and SDcards provide 16, 32, 64, and even 128 GB of storage space. Try to imagine 256 billion of the boards shown below supported in huge steel racks and all powered up. If such a construct could ever be assembled, the entire memory core would be the size of a city with a population of 100,000 people. The addressing system for the core would probably be 1/4 of that size in addition. The processing core would likely fit into a small sports stadium. The power requirements would likely require a few nuclear reactors outputting power at a level of at least 100 gigawatts. Um, we buy that much memory nowadays on a tiny card that costs $50 ? I believe that the engineer from 1972 would say... "Yeah, right, when Hell freezes over."

Plessey Memories Inc. Santa Ana. California

The actual memory array is a real work of art. The label is applied to a sheet of 1/8" thick plexiglass or lexan that is mounted about 3/16" above the memory array. The plastic sheet provides simple physical protection for the array.

The array is composed of rows of tiny ferrite donut-shaped beads. Each bead is about 0.8mm in diameter. There is a system of wires woven through the holes of the beads. Each row has a wire common to the beads in that row. Each column has a wire common to the beads in that column. A third wire (green) passes through all of the beads. This wire is called the 'sense' wire.

The beads are individually magnetized in one of two polarities. The polarity of the magnetic field of a bead determines the state of the bit of data that it stores. The polarity of the magnetic field is 'written' by applying pulses of current through the row and column wires. The 'address', meaning a specific bead, is determined by the specific row and column wire pair that is energized. A specific threshold of current is required to 'flip' the magnetic field of a bead. Where the row and column wires intersect, their combined currents generate a strong enough magnetic field to flip the field of the bead. All of the other beads along the same pair of wires will be exposed to the field current of only one wire which is insufficient to flip their fields and so they remain uneffected.

Okay, so now you know how data (represented by magnetic fields of one polarity or the other) is written into the beads, so how is the data read back? That's where the sense wire comes into play. To read a bit of data, a specific bead (bit) is addressed (by a row and column pair of wires) and the wires are pulsed with current using a certain polarity (lets call this polarity a logic one). During the pulse, if the magnetic field polarity of the bead is flipped, the sense wire will output a pulse. If the bead field doesn't flip, the sense wire outputs no pulse. In other words, if the sense wire pulses, the data contained was of the opposite polarity which was a logic zero. If the sense wire doesn't pulse, the bead field was already in the polarity of a logic one. One drawback of this method of reading the data is that it destroys the logic zero states by replacing them with logic one states. In other words, if you want the original data to be retained, you have to write it back into the memory array after it is read. An advantage of this type of memory is that it is non-volatile meaning that it will retain the data without power for many years. Like any magnetic memory however, it can be entirely erased by direct exposure to any strong external magnetic field.


Radio Shack Toys Circa 1981:

Does anyone remember the remote controlled Flexmobile from Radio Shack? This vehicle had a blinking light on top. I added lamps to the two lenses on the front corners of the roof and to the small lamp on the roof at the rear center. This vehicle was driven by two sticks, one stick for each track providing forward and reverse. I had purchased this with the intent to add a lot of my own electronics to the cargo bay. The entire rear section of the red body was easily removed (simple snap-on/off) to reveal a large empty space that was perfect for add-on circuits. I never actually did anything with it, however, using modern robotics IR sensors, cameras, and microcontroller based controller boards, this vehicle could be made full autonomous. When I bought this item, it was in a scrap bin because someone had dropped it and a gear was stripped so it would not drive one of the tracks. I was able to purchase the full manual for the product from which I looked up the part number for the gear. I then ordered a new gear and installed it. My total cost was something like $15 for a toy that normally cost about $60. Back in those days, you could get the full engineer documentation with mechanical drawings and electronics schematics and you could actually order individual components. Try doing that today and you'll be quite disappointed.


Another relic from Radio Shack, the Z-707 Battle Iron Claw. A tracked vehicle with a remote controled arm that moves up and down and a claw that opens and closes. The big red light on top flashes. This vehicle was fun to drive and could easily pick up and move light objects such as foam blocks.


Radio Shack also sold a variety of RC cars back in the early '80's. One of my favorites was the 911 Porsche with proportional steering and 3-speeds of forward and one reverse. While the chassis and body of these cars was engineered to support lamps, it contained none. The solid design made this toy a prime candidate for adding lights and other circuitry. The car was available in black or white.

During my 2'nd year in college (1983), I was required to design, build, and demonstrate a complex electronic circuit. I chose to build a speed controller with no moving parts for the RC car shown above. We had no computers to work with back then and there was no internet from which to glean knowledge and examples, so this was the result of a lot of library book reading and creative thinking. If I had to do this today, I'd use a small low cost microcontroller IC and some MOSFE Transistors and the entire thing would be about 1/12" the size of this. However, in 1983, there really wasn't anything like that available to students so I had to create everything with a combination of hard logic and analog components.

The speed controller (above) used 555 timer chips as digital timer/counter clocks. The RC receiver would send the speed pulses (0.5,s to 1.5ms pulses repeated at about 20 to 30 Hz) to the controller. The controller used a digital counter to determine the pulse length. The digital length value was then latched (stored) and fed into a simple R2R resistor network to convert it into an analog value. There was also a digital method of detecting whether or not the pulses were within the ranges of forward, reverse, or stop. The analog voltage representing the decoded value for forward or reverse was directed into a comparator circuit. Another 555 timer generated a triangular reference waveform that was also sent into the comparator circuit. The output of the comparator circuit was a fast PWM signal that then controlled an H-bridge that then drove the motor. The PCB was hand drawn, etched in a tray, hand drilled, and then soldered together. This circuit is primitive by today's standards but this was the 2'nd most complex circuit produced within the 3 technology classes at the time. If I were to have to build something like this today, I'm sure that I would be programming a much smaller and smarter circuit as well as routing it on the PC. Considering how easy it is today to put together something vastly superior, it makes me feel like everything that I produce these days is cheating. Everything was so much harder to do back then!

Here's another Radio Shack RC Car Project 1981:

Before I went to college to take the electronics engineering technology program, I picked up a broken RC Porsche 911 toy car from Radio Shack and added a ton of lights and other stuff to it. This was the white version of the toy and I painted it red. I was studying electronics at home and I had been adding lights and motors to static plastic models for a few years and I wanted to assemble and RC car that would demonstrate that such toys could also be fitted with more than what is required to drive around. This car could operate the lights from the batteries or from an external supply so that it could just sit on a shelf (on display) and show off the lights.

I built a circuit board with a 1-of-8 LED sequencer and installed it into the car. The front and rear 5-LED Night Rider displays as well as the two 8-LED circular displays were driven by this logic board. The headlights, markers lights, and taillights were all hardwired to stay on. Signal lights and reverse lights were switched by microswitches mounted near the speed control cam and the steering servo such that those lights operated automatically as the car was driven.

The 1-of-8 LED sequencer board with anti-theft alarm amplifier and piezo speaker (below).

Today, this would be done with a small microcontroller and some simple code. As it is, it's all TTL. First, a 555 timer provides a clock. The clock triggers two D-FlipFlops (7474) that divide the 555 clock twice. The clock plus the two flipflop signals produce a simple 3-bit repeating and sequencial address. The 3-bit address is decoded into 1-of-8 discrete outputs by a 74138 decoder chip. These days, all of this can be accompished within a single chip such as a Microchip PIC16F88 or a Parallax Stamp1 or Stamp2. The 1-of-8 sequence is used to drive the 8-LED arrays in a circle.

The front and rear 5-LED Night Rider displays are sequenced as shown below by wiring two channels to some LEDs (using diodes to OR them together as shown below...

I actually have no recollection of building this weird board (below). I don't recall what it does at all but if I had to guess, I'd say that since it has four 7905 voltage regulators on it, that it provides regulated power to the lighting system. Why does it have a 555 timer on it then? Maybe it was used to flash some lights, the blinkers maybe. The board was probably used to provide multiple functions. I made this thing 31+ years ago so I guess I can forgive myself for forgetting about it. I wish someone had taught me about connector blocks and header pins back then. All those wires soldered into randomly placed PCB holes makes me cringe.

This little board (below) is the heart of my anti-theft device. I was afraid that someone might walk away with the car when my back was turned in the event that I ever had the car on display in a public place so I built an alarm into it. The metal dome shaped components are small mercury switches. They were tilted slightly to put them close to tripping while the car was level. If someone were to pick up the car and tilt it either way, one of the switches would close and set off the alarm. It may have been wise to add another pair of mercury switches mounted 90 degrees to the first pair to protect the other tilt axis.

Well, in 1981, it was pretty time consuming and difficult to build PCB's at home so I often resorted to interconnecting components using the old fashioned point-to-point method (such as was done in very old radios). This pictures demonstrate how I once did things using very little in terms of resources. Obviously, I had time and no money. Today, I have money and no time. I would never assemble anything like this today. Things sure have changed. Below, the circular LED arrays and charging LEDs are shown from the inside. Eww.

Tail end Night Rider display and taillight lamp...




Arcade ---- or ---- Home


Last updated: July 29, 2012

Really Old Electronic Projects: