Computer architecture: Digital Logic - lecture 4b - Charles Fox - University of Lincoln
Watch on YouTube →
Overview
Charles Fox explains how Claude Shannon's switching-circuit ideas combine transistor-based logic gates with George Boole's algebra to build scalable digital systems. The lecture progresses from fabricated gates and combinational components such as multiplexers and adders to clock-controlled feedback, flip-flops, registers, and counters, which provide the memory needed for general computation.
Key takeaways
- Transistor-based logic gates restore signal levels using power from the circuit, which lets many gates be composed without the signal degradation of a bare transistor switch.
- A three-bit decoder can select one of eight output lines, while a multiplexer performs the reverse routing task by forwarding one of several inputs.
- A half adder uses XOR for the sum and AND for the carry; a full adder adds a carry-in and passes carry-out to the next binary column.
- Combinational logic has no memory because its outputs depend only on current inputs; feedback creates persistent state, but must be controlled to avoid unstable behavior.
- Clock edges coordinate state changes, and flip-flops combine into registers and counters that store and update multi-bit values.
- The SkyWater 130 open-source design flow and Logisim export work described in the lecture connect gate-level designs to a path toward physical chip fabrication.
Chapters
- Claude Shannon connected Boolean logic to switching circuits, showing how gates could be used to build computing devices.
- A single transistor can resemble an AND gate, but its output may be degraded by mixed or weakened electrical signals.
- Multi-transistor gates draw on power and restore output voltages, allowing gates to be connected in sequence without losing signal quality.
- Older TTL chips package small collections of gates, often four copies of one gate type, for use in hardware projects such as robotics.
- Modern chip layouts arrange transistors and wiring into standard cells, with power and ground rails running across the chip.
- The SkyWater 130 process provides an open-source chip-design flow; Logisim designs can be carried through open-source tools toward fabrication.
- A contribution from the University of Lincoln modified Logisim to support the design-export pipeline.
- A multi-input AND gate can be assembled from two-input AND gates and turns on only when every input is on.
- A binary shifter moves bits between columns, multiplying or dividing by powers of two with a small number of gate stages.
- A three-bit decoder uses its eight possible input codes to activate one of eight output lines, a useful pattern for selecting CPU operations.
- A multiplexer uses control bits to route one of several input signals to a single output.
- A half adder adds two bits: XOR produces the sum, while AND produces the carry.
- A full adder combines two input bits with a carry-in, using two half adders and an OR gate to produce a sum and carry-out.
- Boolean algebra can rewrite an adder using available components, such as XOR and NAND gates on TTL chips, before laying out the circuit.
- Multi-bit addition chains one adder per column, but each column must wait for the preceding column's carry.
- Combinational circuits such as adders, shifters, decoders, and multiplexers calculate outputs directly from current inputs.
- Because combinational logic has no state, it cannot retain values over time or simulate machines that require memory.
- The lecture connects general computation, described through Church's account of computation, to the need for state and stored information.
- Feedback can cause unwanted effects, illustrated by recursive screen images, bridge resonance, and electric-guitar audio feedback.
- A guitar-and-speaker loop can sustain a sound after a brief string vibration, demonstrating how feedback can preserve a zero or one state.
- Early mercury delay-line memories used microphones, speakers, and acoustic signals traveling through a pipe to retain data.
- Feeding a NOT gate's output back to its input represents the unstable condition X = NOT X, while an OR-gate loop can latch a triggered one.
- A clock, often based on a quartz crystal, generates a regular square wave that coordinates circuit updates.
- Digital circuits commonly use the positive edge, the transition from zero to one, as the moment when state updates.
- An SR latch made from cross-coupled NOR gates uses set and reset inputs to store and clear a bit.
- A D flip-flop captures its input at a clock edge and holds that stored value until a later update.
- A register combines single-bit storage elements into a multi-bit word, such as a four-bit register.
- A counter updates its binary value in response to clock signals, with stored bits providing carry behavior between stages.
- Sequential logic adds controlled feedback and memory to Boolean gate networks, enabling computations that combinational logic alone cannot perform.
- Fox emphasizes that logic gates and a clock together can implement any computable function expressible in logic.
Transcript excerpt
0:00 Claude Shannon had been trying to fix this telephone interconnection problem. And during his MSC thesis, he created the logic gate concept, and he went to the library and he dug out Bull's theory of what is now called Boolean logic to help him simplify designs made from these logic gates. Um So, Shannon's thesis has two aspects. It has the low-level aspect, which is how do you make a logic gate? It has a high-level aspect, which is what can you do with a logic gate, which is building computing devices and applying Boolean logic to them. Um Let's just take a closer look at how the logic gates are actually constructed. So, here here is an example of a NAND
0:49 gate. And you see it is made from transistors. This is the symbol for a transistor. This is the switch. Yeah, as as as we saw earlier, the transistor has three wires coming out. It's got a main current, and you're turning that main current on and off according to a switching signal sent to the base here. Um Now, in in 1936, they didn't have transistors, so Shannon would have been doing this with valves, I think, but they're they're conceptually the same thing. A valve is just a a different electrical instantiation of the switching idea. Nowadays, we can just take Shannon's theory and run it directly with transistors, including the transistors that are being fabricated um onto these pieces of silicon.
1:37 Um Students often ask, why do you need this level of complexity when you could, for example, use a single transistor? It it occurs to many students that a single transistor actually looks a lot like an AND gate. Because if you if you have your main current flowing through here, that's like one of the inputs. And your second
This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.
Related notes
Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, Charles Fox.