This report covers the simulation (and comparison) of 3 of the most commonly used MOS current mirror topologies (refer to pages 3-7 and 3.8 from our class textbook below), mainly:

  1. Book figure 3-23 (widely used mirror)
  2. Book figure 3-24 (fewer devices, same performance)
  3. Book figure 3-25 (best performance)

Schematic Diagram

1. Widely Used Mirror

Widely Used MOS Current Mirrors 1 Schematic

2. Fewer Devices, Same Performance

Widely Used MOS Current Mirrors 2 Schematic

3. Best Performance

Widely Used MOS Current Mirrors 3 Schematic

SPICE Simulations

For the SPICE simulation code, see “Source Code” section below.

Operating Point Analysis

Here, we are calculating the DC voltages (bias voltages) at every node of our circuits.

Note: Node DC measurements (re-formatted for display).

1. Widely Used Mirror

Node                    Measurements
----                    ------------
n1                      1.1692 V
n2                      1.0 V
n3                      1.1786 V
n4                      .4162 V
n5                      .4166 V
n_pos                   3.0
v1#branch               -100.0 uA
v2#branch               -49.9838 uA
(v2#branch/(-50e-06))   .9997

2. Fewer Devices, Same Performance

Node                    Measurements
----                    ------------
n1                      1.1711 V
n2                      1.0 V
n3                      .4108 V
n4                      .4104 V
n_pos                   3.0 V
v1#branch               -50.0 uA
v2#branch               -49.9827 uA
(v2#branch/(-50e-06))   .9997

3. Best Performance

Node                    Measurements
----                    ------------
n1                      .9555 V
n2                      1.0 V
n3                      1.2055 V
n4                      .4371 V
n5                      .4373 V
n_pos                   3.0 V
v1#branch               -50.0 uA
v2#branch               -50.0007 uA
(v2#branch/(-50e-06))   1.0000

DC Analysis

In our DC analysis, we are measuring the variation of the mirrored output current under different loads.

We are applying a DC sweep to V2 (our load voltage) from 0 to 3V in 0.1V increments.

We are plotting the output current magnitude vs drain voltage. (our load voltage at n2)

1. Widely Used Mirror

Widely Used MOS Current Mirrors 1 Simulation DC

2. Fewer Devices, Same Performance

Widely Used MOS Current Mirrors 1 Simulation DC

3. Best Performance

Widely Used MOS Current Mirrors 1 Simulation DC

Results

This lab explores the simulation of 3 common MOS current mirror topologies.

As noted in the book (refer to pages 3-7 and 3-8 below), we can see that a common technique used for all the circuits is the addition of a cascode stage at the output. This stage helps to shield the lower current matching transistors from fluctuations in load voltage and reduces the output current dependence on voltage. (improving our output resistance figure)

Biasing for the cascode stage transistors (required to keep them in saturation where drain current dependence on drain voltage is small) is provided by both the use of additional devices (e.g. M1 for circuit 1 and R1 for circuit 3) and by changing the dimensions of MOS devices, i.e. by changing the dimensions of MOS devices one can affect their threshold voltage indirectly. (V_th is actually dependent on many other physical and process parameters which vary depending on the semiconductor technology being used)

Furthermore, we can see that there are additional transistors (M1 for both circuits 2 and 3, and M2 for circuit 1) added in order to bring the drain voltages of our lower current matching transistors more in-line with each other to further reduce our current match error.

This report concludes our simulations for chapter 3 of the book – current mirrors – upcoming reports will explore “Differential Pair” IC topologies with bipolar and MOS devices.

Error Measurements

1. Widely Used Mirror

For the first circuit, we have a variation of 49.9617uA to 50.1523uA over an operating range of 0.8V to 3V. This is equivalent to an error of 0.3812% relative to the current reference.

2. Fewer Devices, Same Performance

For the second circuit, we have a variation of 49.9594uA to 50.1617uA over an operating range of 0.8V to 3V. This is equivalent to an error of 0.4046% relative to the current reference.

3. Best Performance

For the third circuit, we have a variation of 49.9947uA to 50.0256uA over an operating range of 0.7V to 3V. This is equivalent to an error of 0.0618% relative to the current reference.

Figures of Merit

1. Widely Used Mirror

  • Output Resistance: 11.5425MR (from 0.8 to 3V linear range)

  • Compliance Voltage: 0.8V (from ground)

2. Fewer Devices, Same Performance

  • Output Resistance: 10.8749MR (from 0.8 to 3V linear range)

  • Compliance Voltage: 0.8V (from ground)

3. Best Performance

  • Output Resistance: 74.4337MR (from 0.7 to 3V linear range)

  • Compliance Voltage: 0.7V (from ground)

Looking at our results above, we can see that indeed the second circuit is comparable in performance to the first circuit (while employing fewer transistors and references) as described in the book, and the third circuit is evidently superior in both output resistance and compliance voltage and presents the best performance of the 3 topologies.

Thus our simulations agree with our reference book results and expectations, and that is always a great thing.

Source Code

References



Leave your comments below (or comment directly here).

Thank you for your feedback.