Apollo-Era Digital Logic

I was in High School when the Apollo space program was in full swing.  I was just beginning to get interested in computers, but had only a vague idea about how they worked.  I think one of the things that caught my eye and led to this interest in computers was the Apollo Guidance Computer (AGC) which in many ways, was one one of the key elements that led to a successful moon landing.  If you're interested, there's a wonderful old documentary on YouTube called "Computer for Apollo" which shows the development and construction process of the AGC at the MIT Instrumentation Laboratory and the Raytheon Corporation.  The construction technique shown in the film must be for an early version of the AGC, as it shows the logic "sticks" being built using a pre-PCB process using ICs packaged in modified transistor cans.  I believe that later versions of the AGC switched to using an early type of PCB and ICs in flat packages.

Click for larger view

Recently I saw a documentary on the early days of Silicon Valley that focused on the beginnings of Fairchild Semiconductor, which talked briefly about some of the first integrated logics circuits produced by Fairchild and which were used as the basis of the AGC.  So, I wondered if if might be possible to track down some of these early "micro logic" devices and try them out.  The AGC was reportedly built from a single type of logic circuit, a dual, triple NOR gate, but I was unable to located similar devices on the surplus market.  However, after a bit of searching, I found a Canadian seller on eBay that sold me a set of 5 uLogic 923 devices which function as JK Flop Flops.  The uL923 is from them same family of devices used in the AGC which are based on a type of logic which Fairchild called RTL, for Resistor Transistor Logic.  Unlike the ICs we use today, these early devices where packaged in TO-99 metal cans, or pin-indentical epoxy versions.  The uL923 comes in an 8 pin package with radial leads, so I decided to use 8 pin transistor sockets and design a single adapter board so I could easy mount them on a conventional breadboard while keeping the lead wires in their original, unbent position.  Here's a uL923 mounted on one of my socket adapters:

Click for larger view

I located a scanned version of the data sheet for the uL932 using a web search.  based on this, I decided to wire up a simple, 5 bit ripple counter as a demonstration of a working subsystem and as way to investigate how fast these device are capable of switching.  Here's the schematic for the circuit I wired up:

And, here's a photo showing the same circuit wired up using a conventional breadboard:

Click for larger view

The data sheet indicated a working voltage of 3.6 volts for the device, so that is what I used.  I started by feeding in a 2 MHz square wave as the clock signal, as the data sheet indicated that the device was guaranteed to work at this frequency.  I then set the amplitude of the clock input at 1.5 volts after observing that the output swing from the first stage was about 1.5 volts.  Miraculously, the circuit worked after correcting only a few misfires and all 5 devices seemed to function producing outputs that divided down to clock/32.  I then experimented with increasing the clock rate and was able to bump it up to 22 MHz before the first stage flip flop started to misbehave.  The reported clock speed of the AGC was 1.024 MHz, so it looks like the designers had a good safety margin in the uLogic devices.  Counting at 2 Mhz, my ripple counter circuit draws only 100 mA and this does not increase as the clock speed goes up to 22 MHz.  Here's a video that shows the ripple counter in operation: