Lecture 16: Variants of Logic Circuits - Part 1
Overview
The lecture compares CMOS with ratioed logic, differential cascode voltage switch logic (DCVSL), and pass-transistor logic, showing how each trades transistor area against power, voltage swing, robustness, and design complexity. It works through ratioed-logic sizing, cross-coupled DCVSL for complementary outputs, and compact pass-transistor XOR/XNOR examples, including why CMOS remains the general-purpose choice despite higher transistor counts.
Key takeaways
- Ratioed logic saves the complementary PMOS network, but its low output is set by a resistance divider and its always-on pull-up draws steady-state current during pull-down.
- DCVSL uses cross-coupled PMOS gates and complementary NMOS networks to avoid pull-up/pull-down contention while producing complementary outputs together.
- The optimized DCVSL XOR/XNOR pair uses eight transistors, while the lecture's CMOS XOR example uses ten including an input inverter.
- An NMOS pass transistor passes a degraded high, roughly VDD minus its threshold voltage, so cascaded gate-controlled stages can accumulate voltage loss.
- Pass-transistor logic can reduce transistor count substantially, but CMOS retains advantages in rail-to-rail swing, noise recovery, broad function support, and design automation.
Chapters
- The lecture moves from conventional CMOS logic, adders, and multipliers to alternative circuit families.
- Some specialized designs prioritize area reduction enough to revisit older technologies, even though CMOS is the course's main design platform.
- A conventional CMOS gate generally uses complementary NMOS and PMOS networks, increasing transistor count compared with some alternatives.
- Ratioed logic keeps the designed NMOS pull-down network but replaces the complementary PMOS network with an unconditional pull-up load.
- A PMOS with its gate tied to ground stays on, so the NMOS network conditionally pulls the output toward ground.
- The reduced transistor count comes with continuous current when the NMOS network is active, unlike static CMOS's disconnected steady-state path.
- When both networks conduct, the output is set by a voltage divider: approximately Vout = RN/(RN + RP) × VDD.
- A higher PMOS resistance RP lowers the output's low-state voltage; using a minimum-width PMOS raises resistance.
- Because the output level depends on the PMOS-to-NMOS resistance ratio, ratioed logic may not achieve rail-to-rail swing.
- The example implements ABC + DE + G as an NMOS pull-down network with three parallel branches.
- The ABC and DE branches use series NMOS transistors, while G forms its own branch to ground.
- A single always-on PMOS load connects the output toward VDD instead of a CMOS pull-up network.
- Differential cascode voltage switch logic (DCVSL) pairs a function with its complement, such as AND and NAND.
- Each output drives the PMOS gate in the opposite logic cell, turning that pull-up off when its output should be low.
- Unlike ratioed logic, the cross-coupled arrangement avoids a continuously conducting pull-up versus pull-down fight.
- The XOR function is A B̅ + A̅ B, while XNOR is A B + A̅ B̅.
- Separate NMOS pull-down networks implement the complementary functions, and their outputs control the opposite PMOS loads.
- The paired circuit supplies both XOR and XNOR, which is useful when a design needs a function and its complement.
- The optimized DCVSL XOR/XNOR design shares input-dependent NMOS branches between the two complementary functions.
- The resulting pair uses eight transistors, compared with ten for a single CMOS XOR including its input inverter.
- CMOS remains easier to automate and generalize; DCVSL's area advantage depends on needing complementary outputs and suitable shared logic.
- Pass-transistor logic differs from CMOS and DCVSL by applying data inputs to transistor drains or sources as well as to gates.
- A transistor controlled by B can pass A to the output when B is high, implementing the selected A·B case.
- A complementary control branch is needed to define the output when the first transistor is off, rather than leaving it disconnected.
- An NMOS pass transistor does not transmit a full VDD high level, producing an output below the supply voltage.
- A CMOS inverter can restore a degraded signal to a rail, but a second inverter may be required if the original polarity must be preserved.
- Adding restoration circuitry can reduce the transistor-count advantage that motivated pass-transistor logic.
- A high passed through an NMOS can fall to roughly VDD minus the transistor threshold voltage.
- Passing the degraded result into another gate-controlled stage can subtract another threshold voltage, weakening later logic stages.
- Pass-transistor outputs are therefore harder to branch and cascade than CMOS's rail-to-rail signals, which support long logic pipelines.
- With both true and complemented inputs available, the illustrated pass-transistor arrangement forms XOR from A B̅ + A̅ B and XNOR from A B + A̅ B̅.
- The complementary pair uses four pass transistors in the example, half the eight-transistor optimized DCVSL pair before output restoration.
- The topology is less robust to noise and requires more manual Boolean reasoning than CMOS, especially for complex functions.
- Pass-transistor NAND design needs branches that pass a defined zero when a control transistor disconnects the other path.
- The lecture applies Boolean algebra and De Morgan's theorem to explain why a less-obvious branch combination implements NAND.
- A feedback loop using a PMOS is introduced as a way to restore degraded output levels; extra restoration transistors remain worthwhile when larger logic blocks save enough area.
Transcript excerpt
0:03 All right. So today we'll talk about other variants of logic circuits. So far you must have gone through the conventional SMOS logic design approaches and then we followed it up with the error design which is a specialized SIMOS logic circuit you know with a purpose and then you must have seen the multiplier design which basically uses the adder many times over. So I hope you see the connections there. We build upon simple SMOS inverter and then use the same principle to design generalized SMOS logic circuits. Then use that principle to design an adder and then use in I don't know it's like doing this for some I have a okay hopefully it stays that
0:53 way. So uh yeah so there is a levels of abstraction uh that you generally can see. So today I I [laughter] have a feeling that we have had enough practice with logic design. Uh we were ready to see what's beyond Simos. So I will remind you most of these technology that we discussed today they came before SIMOS. So you might be wondering why do we even care? I mean we have already learned the most cutting edge. That's why we started there. We wanted to focus most of our energy in the SMOS stuff. That's why the project is on SMOS. All the labs you're doing are on SMOS, right? But remember, every once in a while for specialized applications, we often revive some of these seemingly older logic families. Imagine you have a specialized
1:40 application which absolutely require minimizing area and minimizing area beyond the SIMOS stuff. One major criticism about CMOS and I'm going to get started with that criticism first of all. You see, if you need 10 NOS transistors, you will need 10 POS transistors as well for
This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.
In this course
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, NorDIC Lab.