What prevents this transistor based discrete logic gate design from having been used?

Question

I am interested in building discrete logic gates as a teaching / learning tool to explore computer fundamentals. The easiest logic gate design I can come up with consists of the following:

It is simple enough to be built from discrete THT transistors, and Ben Eater, for example, uses this type of design when exploring logic gate fundamentals.

However, this does not seem to have been used in industry, as I cannot find a (named) design technique that uses these. Thus, I assume it has serious problems when used in practice, either alone or when combining several to build a more complex circuit, such as a register or an adder.

Q: What would be the main problems in this design technique that prevent it from being used for more complex designs than a single discrete gate?

As a bonus question, would you avoid this technique as a learning / teaching tool when exploring logic gate fundamentals?

Answer 1

Saturating logic is slow.

Resistor based logic is slow, fairly power hungry, and not easy to integrate into ICs.

Looking at the development history, the first type of 'standard' logic was RTL, resistor transistor logic. That was then replaced by DTL, diode transistor logic. However, when trying to integrate multiple diodes with transistors, the multi-emitter transistor was found to be far easier and smaller to make, and NAND TTL was born.

It's a good tool to teach logic, as long as you're prepared to go the extra mile with the brightest in the class who already know about TTL and CMOS.

It's worth noting that anything that can do boolean operations can be used to create logic. Amongst those are relays, flows of water (fluidics), persistent patterns on Conway's 'Life'. We only use CMOS dominantly now as it's higher speed and cheaper to integrate than most of the other possible implementations.

Answer 2

OP's RTL logic is certainly simple. It is unavoidable that basic bipolar transistor characteristics must be well-understood first.
You might consider substituting discrete N-channel MOSFETS rather than bipolar transistors. RTL logic gates using MOSFETS were not at all common in the historical record, but experiments with basic N-channel MOSfet gates might ease the transition to modern complementary MOS logic that uses both N-channel and P-channel MOSFETs.

Those proposed simple logic gates like NAND, NOR, NOT might be useful for illustrating the following aspects of logic interface, since (as Neil has pointed out) they are slow:

• have Limited fanout
• different Rise time from Fall time
• BJT storage time
• noise immunity limited by fanout

^{simulate this circuit – Schematic created using CircuitLab}

MOSFETs logic is simpler, not needing a resistor on the GATE, unlike BJT gates. And threshold voltage of MOSFETS is higher than threshold for BJT. But the fanout limitation is better illustrated by the BJT arrangement.

It may be a bit risky leaving a MOSFET gate open-circuit as illustrated. The N-channel MOSFET NAND might be safer with large-value pull-down resistors to GND on each input. It would emphasize the danger of leaving modern CMOS logic gates with floating inputs.

Answer 3

Basic idea

It would be interesting to explain the reason for the low speed of this type of logic gate (RTL), which led to their replacement with TTL. In my opinion, it is the presence of a base resistor connecting the output of the previous stage to the input (BJT base) of the next stage.

RTL vs TTL

It is interesting to examine and compare the input parts of the two types of logic gates families because they differ greatly in the way they are controlled.

RTL

Indeed, a base resistor Rb is required at the input of the BJT to limit the current. The problem here, however, is that the transistor is switched on and off through the same resistor.

Vin = HIGH: The base is charged by the output of the previous stage through Rb.

^{simulate this circuit – Schematic created using CircuitLab}

Vin = LOW: The base (slowly) discharges through Rb and the bottom (turned on) transistor of the previous output stage.

^{simulate this circuit}

TTL

In TTL only switching on is through a "pull-up" base resistor connected to the own power supply.

Vin = HIGH: The Q2 base is charged through Rb and the Q1's forward-biased base-collector junction (diode). The interesting thing here is that the input voltage source has a voltage but it does not drive Q2; the supply voltage does it through Rb. All this complexity is done in order to be able to connect the Q2's base to the ground directly and not through Rb when Vin = 0 (see below).

^{simulate this circuit}

The Q1 base-emitter junction acts as an open switch that switches off the input voltage.

^{simulate this circuit}

Vin = LOW: The switching off is done by shorting the input base-emitter junction. So the base (quickly) discharges through the Q1's collector-emitter part and the bottom (turned on) transistor of the previous output stage.

^{simulate this circuit}

The Q1's collector-emitter part acts as a closed switch that grounds the input (Q2's base).

^{simulate this circuit}

In this way, the transistor turns off quickly because the charge on its base is dissipated quickly.