Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

1 Existence and Uniqueness of Solutions To Differential Equations

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

EECS 16B Spring 2020 HW 1

1 Existence and uniqueness of solutions to differential


equations
Let’s show that if any function 𝑥 satisfies
d
𝑥(𝑡) = 𝛼𝑥(𝑡) (1)
d𝑡
as well as

𝑥(0) = 𝑥 0 , (2)

then it is unique: if 𝑦 is any function that meets these two criteria then 𝑥 = 𝑦.
In order to do this, we will first verify that a solution exists. Then we
will compare it to a hypothetical alternative solution—and our goal will to be
establish that these two solutions are equal.

a) Verify that 𝑥 𝑑 (𝑡) = 𝑥0 𝑒 𝛼𝑡 satisfies (1) and (2). (For this proof, 𝑥 𝑑 will be
the “reference solution” against which alternates will be compared.)

Solution
Taking the derivative of 𝑥 𝑑 (𝑡) with respect to 𝑡 gives 𝛼𝑥0 𝑒 𝛼𝑡 by the chain
rule, and this is equal to 𝛼𝑥 𝑑 (𝑡) by inspection. So (1) is satisfied.
Evaluating 𝑥 𝑑 (0) = 𝑥0 𝑒 𝛼·0 = 𝑥0 and so (2) is also satisfied.
b) To show that this solution is in fact unique, we need to consider a
hypothetical 𝑦(𝑡) that also satisfies (1) and (2).
Our goal is to show that 𝑦(𝑡) = 𝑥(𝑡) for all 𝑡 ≥ 0. (The domain 𝑡 ≥ 0 is
where we have defined the conditions (1) and (2). Outside of that domain,
we don’t have any constraints. )
How can we show that two things are equal? In the past, you have
probably shown that two quantities or functions are equal by starting
with one of them, and then manipulating the expression for it using valid
substitutions and simplifications until you get the expression for the other
one. However, here, we don’t have an expression for 𝑦(𝑡) so that style of
approach won’t work.
In such cases, we basically have a couple of basic ways of showing that
two things are the same.
• Take the difference of them, and somehow argue that it is 0.
• Take the ratio of them, and somehow argue that it is 1.
We will follow the ratio approach in this problem. First assume that
𝑦(𝑡)
𝑥0 ≠ 0. In this case, we are free to define 𝑧(𝑡) = 𝑥 𝑑 (𝑡)
since we are dividing
by something other than zero.
What is 𝑧(0)?

1
HW 1 @ 2020-02-11 17:16:14-08:00

Solution
𝑦(0)
We know 𝑧(0) = 𝑥 𝑑 (0) = 𝑥𝑥00 = 1 since 𝑦(0) = 𝑥 0 by (2) and plugging in 0
for 𝑡 into the exact expression for 𝑥 𝑑 (𝑡).

c) Take the derivative 𝑑𝑡𝑑 𝑧(𝑡) and simplify using (1) and what you know
about the derivative of 𝑥 𝑑 (𝑡).
(HINT: The quotient rule for differentiation might be helpful since a ratio is
involved.)

Solution
𝑦
The quotient rule tells us how to take the derivative of 𝑥 𝑑 (we can also
view this using the product rule, which is also just another manifestation
of the chain rule for differentiation in the multivariate case). The rule
applies because the functions involved are differentiable by definition
and the denominator is nonzero.

𝑑
𝑑 𝑑 𝑦(𝑡) 𝑑𝑡 𝑦(𝑡)𝑥 𝑑 (𝑡) − 𝑦(𝑡) 𝑑𝑡𝑑 𝑥 𝑑 (𝑡)
𝑧(𝑡) = = (3)
𝑑𝑡 𝑑𝑡 𝑥 𝑑 (𝑡) (𝑥 𝑑 (𝑡))2
𝛼𝑦(𝑡)𝑥 𝑑 (𝑡) − 𝑦(𝑡)𝛼𝑥 𝑑 (𝑡)
= (4)
(𝑥 𝑑 (𝑡))2
0
= =0 (5)
(𝑥 𝑑 (𝑡))2

Notice that here, what is important is that both 𝑦 and 𝑥 𝑑 satisfy (1) and
so the numerator in the quotient rule cancels out to zero. The details of
𝑥 𝑑 (𝑡) didn’t end up mattering.
You should see that this derivative is always 0 and hence 𝑧(𝑡) does not
change. What does that imply for 𝑦 and 𝑥 𝑑 ?

Solution
Since 𝑧(𝑡) has zero derivative, it cannot change, and hence it stays at its
𝑦(𝑡)
initial value, which is 1. So it is always 1 and hence 𝑥 𝑑 (𝑡)
= 1 so 𝑦(𝑡) = 𝑥 𝑑 (𝑡).

d) At this point, we have shown uniqueness in most cases. Just one special
case is left: 𝑥 0 = 0. The ratio technique omitted this case, because as
𝑥 𝑑 (𝑡) = 0, 𝑥 𝑑 cannot be the denominator of a fraction.
To complete our proof we must to show that if 𝑥0 = 0, then 𝑦(𝑡) = 0
for all 𝑡, and we will do so by assuming that 𝑦(𝑡) is not identically 0 for
𝑡 > 0—that is, at some 𝑡0 > 0 𝑦(𝑡0 ) = 𝑘 ≠ 0.

2
EECS 16B Spring 2020 HW 1

From (2), we know that 𝑦(0) = 0. In this part, we will try to work
backwards in time from the point 𝑡 = 𝑡0 to 𝑡 = 0 and conclude that 𝑦
violates (2).
Apply the change of variables 𝑡 = 𝑡0 − 𝜏 to (1) to get a new differential
𝑑
equation for e 𝑥 (𝜏) = 𝑥(𝑡0 − 𝜏) that specifies how 𝑑𝜏 𝑥 (𝜏) must relate to
e
𝑥 (𝜏). This should hold for −∞ < 𝜏 ≤ 𝑡0 .
e

Solution

𝑑 𝑑
𝑥 (𝜏) = 𝑥(𝑡0 − 𝜏) (6)
𝑑𝜏 𝑑𝜏
e
= −𝛼𝑥(𝑡0 − 𝜏) (7)
𝑥 (𝜏)
= −𝛼 e (8)

where the second line used the chain rule for differentiation and (1).
This holds for all 𝑡 ≥ 0 which means 𝑡0 − 𝜏 ≥ 0 which is the same as 𝜏 ≤ 𝑡0 .

e) Because the previous part resulted in a differential equation of a form for


which we have already proved uniqueness for the case of nonzero initial
condition, and since e 𝑦 (0) = 𝑦(𝑡0 ) = 𝑘 ≠ 0, we know what e 𝑦 (𝜏) must be.
Write the expressions for e 𝑦 (𝜏) for 𝜏 ∈ [0, 𝑡0 ] and what that implies for
𝑦(𝑡) for 𝑡 ∈ [0, 𝑡0 ].

Solution
𝑑
We know that e 𝑦 satisfies 𝑑𝜏 𝑦 (𝜏) = −𝛼 e
e 𝑦 (𝜏) and that e
𝑦 (0) = 𝑘 ≠ 0. Conse-
quently, by the uniqueness theorem already proved, we know that it must
𝑦 (𝜏) = 𝑘𝑒 −𝛼𝜏 for the range 𝜏 > 0 as long as the differential
be the case that e
equation is valid.
This means that 𝑦(𝑡) = 𝑘𝑒 −𝛼(𝑡0 −𝑡) as long as 0 ≤ 𝑡 ≤ 𝑡0 .

f) Evaluate 𝑦(0) and argue that this is a contradiction for the specified
initial condition (2).

Solution
Evaluating this expression at 𝑡 = 0 gives 𝑦(0) = 𝑘𝑒 −𝛼𝑡0 . Because 𝑘 ≠ 0,
this means 𝑦(0) ≠ 0. This is a contradiction with (2) since that asserts a
zero initial condition 𝑥0 = 0.
Consequently, such a 𝑦(𝑡) cannot exist and only the all zero solution is
permitted — establishing uniqueness in this case of 𝑥0 = 0 as well.

g) Explain in your own words why it matters that solutions to these differ-
ential equations are unique.

3
HW 1 @ 2020-02-11 17:16:14-08:00

Solution
Uniqueness means that we need not continue looking once a satisfactory
guess has been reached. If our model had non-unique solutions—such as
kinematics problem where the quantity of interest arises as a root of a
quadratic—then we would potentially have to choose between multiple,
or even infinitely many solutions.

Although we gave you lots of guidance in this problem, we hope that you
can internalize this way of thinking.
This elementary approach to proving the uniqueness of solutions to differ-
ential equations works for the kinds of linear differential equations that we will
tend to encounter in EE16B. For more complicated nonlinear differential equa-
tions, further conditions are required for uniqueness (appropriate continuity
and differentiability) and proofs can be found in upper-division mathematics
courses on differential equations when you study the Picard-Lindelöf theorem.
(It involves looking at the magnitude of the difference of the two hypothetical
solutions and showing this has to be arbitrarily small and hence zero. However,
the basic elementary case we have established here can be viewed as a building
block — the quotient rule gets invoked in the appropriate place, etc. The
additional ingredients that are out-of-scope for lower-division courses are
fixed-point theorems — which you can think of as more general siblings of the
intermediate-value theorem you saw in basic calculus.)

2 Digital-Analog Converter
A digital-analog converter (DAC) is a circuit for converting a digital representa-
tion of a number (binary) into a corresponding analog voltage. In this problem,
we will consider a DAC made out of resistors only (resistive DAC) called the
𝑅-2𝑅 ladder. Here is the circuit for a 3-bit resistive DAC.

𝑅 𝑅
𝑉out

2𝑅 2𝑅 2𝑅 2𝑅

+ + +
𝑉0 𝑉1 𝑉2
− − −

LSB MSB

Let 𝑏0 , 𝑏1 , 𝑏2 = {0, 1} (that is, either 1 or 0), and let the voltage sources
𝑉0 = 𝑏 0𝑉DD , 𝑉1 = 𝑏1𝑉DD , 𝑉2 = 𝑏2 𝑉DD , where 𝑉DD is the supply voltage.

4
EECS 16B Spring 2020 HW 1

As you may have noticed, (𝑏2 , 𝑏1 , 𝑏0 ) represents a 3-bit binary (unsigned)


number where each of 𝑏 𝑖 is a binary bit. We will now analyze how this converter
functions.

a) If 𝑏2 , 𝑏1 , 𝑏0 = 1, 0, 0, what is 𝑉out ? Express your answer in terms of 𝑉DD .

Solution
There are several ways to solve this problem. For this solution set, we are
going to solve for the generic solution rather than solve for each specific
case of (a), (b), (c), and (d).

𝑉1 𝑅 𝑉2 𝑅
𝑉out

2𝑅 2𝑅 2𝑅 2𝑅

+ + +
𝑏0𝑉DD 𝑏1𝑉DD 𝑏 2𝑉DD
− − −

Applying KCL at nodes 𝑉1 , 𝑉2 , and 𝑉out , we get

𝑉1 𝑉1 − 𝑏0𝑉DD 𝑉1 − 𝑉2
+ + =0
2𝑅 2𝑅 𝑅
𝑉2 − 𝑏 1𝑉DD 𝑉2 − 𝑉1 𝑉2 − 𝑉out
+ + =0
2𝑅 𝑅 𝑅
𝑉out − 𝑏2 𝑉DD 𝑉out − 𝑉2
+ =0
2𝑅 𝑅

Solving this system of equations leads to

𝑏 2𝑉DD 𝑏 1𝑉DD 𝑏 0𝑉DD


+ + = 𝑉out
2 4 8

Plugging in 1, 0, 0 gives the answer.

𝑉DD
𝑉out =
2

b) If 𝑏2 , 𝑏1 , 𝑏0 = 0, 1, 0, what is 𝑉out ? Express your answer in terms of 𝑉DD .

5
HW 1 @ 2020-02-11 17:16:14-08:00

Solution
Plugging into the equation from part (a), we get

𝑉DD
𝑉out = .
4

c) If 𝑏2 , 𝑏1 , 𝑏0 = 0, 0, 1, what is 𝑉out ? Express your answer in terms of 𝑉DD .

Solution
Plugging into the equation from part (a), we get

𝑉DD
𝑉out = .
8

d) If 𝑏2 , 𝑏1 , 𝑏0 = 1, 1, 1, what is 𝑉out ? Express your answer in terms of 𝑉DD .

Solution
Plugging into the equation from part (a), we get

7𝑉DD
𝑉out = .
8

e) Finally, solve for 𝑉out in terms of 𝑉DD and the binary bits 𝑏2 , 𝑏1 , 𝑏0 .

Solution
From part (a),
𝑏 2𝑉DD 𝑏1𝑉DD 𝑏0𝑉DD
+ + = 𝑉out .
2 4 8

f) Explain how your results above show that the resistive DAC converts the
3-bit binary number (𝑏2 , 𝑏1 , 𝑏0 ) to the output analog voltage 𝑉out .

Solution
Every increment of 18 𝑉DD on 𝑉DD represents an increment of 1 to the 3-bit
binary number (𝑏2 𝑏 1 𝑏0 ).
For example, if 𝑉out = 58 𝑉DD , the input was 5 in binary (1 0 1)→ (𝑏 2 = 1
𝑏 1 = 0 𝑏 0 = 1).

6
EECS 16B Spring 2020 HW 1

3 Transistor Switch Model

We can improve our resistor-switch model of the transistor by adding in a gate


capacitance. In this model, the gate capacitance 𝐶 𝐺 represents the lumped
physical capacitance present on the gate node of all transistor devices. This
capacitance is important as it determines the delay of a transistor logic chain.

𝐷 𝑉𝑜𝑢𝑡 𝑆

𝑅 𝑜𝑛 𝐶𝐺 𝑅 𝑜𝑛

𝐼𝐷 𝐼𝐷
𝐺 𝐺

𝐶𝐺 𝑉𝐺𝑆 ≥ 𝑉𝑡𝑛 𝑉𝐺𝑆 ≤ −|𝑉𝑡𝑝 |


𝑉𝑜𝑢𝑡
𝑆 𝐷
(b) PMOS Transistor Resistor-
(a) NMOS Transistor Resistor- switch-capacitor model. Note we
switch-capacitor model have drawn this so that it aligns
with the inverter.

You have two CMOS inverters made from NMOS and PMOS devices. Both
NMOS and PMOS devices have an “on resistance” of 𝑅 𝑜𝑛 = 1 kΩ, and each has
a gate capacitance (input capacitance) of 𝐶 𝐺 = 1fF (femto-Farads = 10−15 ). We
assume the “off resistance” (the resistance when the transistor is off) is infinite
(i.e. , the transistor acts as an open circuit when off). The supply voltage 𝑉𝐷𝐷
is 1V. The two inverters are connected in series, with the output of the first
inverter driving the input of the second inverter (fig. 8).

𝑉𝐷𝐷 𝑉𝐷𝐷

𝑉𝑜𝑢𝑡,1 𝑉𝑖𝑛,2
𝑉𝑖𝑛 𝑉𝑜𝑢𝑡,2

Figure 2: CMOS Inverter chain

7
HW 1 @ 2020-02-11 17:16:14-08:00

𝑉𝐷𝐷 𝑉𝐷𝐷

𝑉𝑖𝑛 low 𝑉𝑖𝑛 low

𝐶 𝐶
𝑅 𝑜𝑛 𝑅 𝑜𝑛

𝑉𝑜𝑢𝑡 𝑉𝑜𝑢𝑡
𝑉𝑖𝑛 𝑉𝑖𝑛

𝑅 𝑜𝑛 𝑅 𝑜𝑛

𝐼𝐷 𝐼𝐷
𝐶 𝐶

𝑉𝑖𝑛 high 𝑉𝑖𝑛 high

Figure 3: Inverter Transistor Resistor-switch model

a) Assume the input to the first inverter has been low (𝑉𝑖𝑛 = 0 V) for a
long time, and then switches at time 𝑡 = 0 to high (𝑉𝑖𝑛 = 𝑉𝐷𝐷 ). Draw
a simple RC circuit and write a differential equation describing the
output voltage of the first inverter (𝑉𝑜𝑢𝑡,1 ) for time 𝑡 ≥ 0. Don’t forget
that the second inverter is “loading” the output of the first inverter — you
need to think about both of them.

Solution
To analyze this circuit as an RC circuit we can recall the transistor switch
model. Using this we can see that the first inverter’s output appears as
a resistor connected to 𝑉𝐷𝐷 when the input is low (nmos off, pmos on),
or a resistor connected to ground when the input turns high (nmos on,
pmos off).
Before 𝑡 = 0, the input to the first inverter was low for a long time. This
means that for 𝑡 < 0, the output of the inverter (𝑉𝑜𝑢𝑡,1 ) had been held at
𝑉𝐷𝐷 for a long time.
At 𝑡 = 0, the input goes high, which means that the input inverter’s nmos
device turns on, connecting 𝑉𝑜𝑢𝑡,1 to ground through a resistance of 𝑅 𝑜𝑛 .

8
EECS 16B Spring 2020 HW 1

The second inverter “loads” the output of the first inverter. From the notes
in the problem, we can model the gates of the transistors as capacitors.
These gates together form our capacitive load. The gate of the pmos acts
as a capacitor to 𝑉𝐷𝐷 and the gate of the nmos acts as a capacitor to ground.

Using this we can draw the following RC circuit:

𝑉𝐷𝐷

𝐶 𝑝𝑚𝑜𝑠 = 𝐶 𝐺

𝑉𝑜𝑢𝑡,1

𝐶 𝑛𝑚𝑜𝑠 = 𝐶 𝐺
𝑅 𝑜𝑛

Figure 4: First inverter output at 0

To get the differential equation describing the output of the first inverter
at time 𝑡 ≥ 0 let us first think about the behavior of the circuit at and after
𝑡 = 0.

Before 𝑡 = 0 we know that the output 𝑉𝑜𝑢𝑡,1 = 𝑉𝐷𝐷 . This means that 𝐶 𝑛𝑚𝑜𝑠
is charged, while 𝐶 𝑝𝑚𝑜𝑠 is not as there is no voltage difference across it.

At 𝑡 = 0, when the input to the first inverter changes (input switches to


high), the nmos will turn on, discharging the 𝑉𝑜𝑢𝑡,1 node. Thus 𝑉𝑜𝑢𝑡,1 will
eventually discharge to zero in steady state.

We know the voltage across 𝐶 𝑝𝑚𝑜𝑠 is 𝑉𝑜𝑢𝑡,1 (𝑡) −𝑉𝐷𝐷 and the voltage across
𝐶 𝑛𝑚𝑜𝑠 is 𝑉𝑜𝑢𝑡,1 (𝑡). Using this information we can set up a differential
equation to solve for 𝑉𝑜𝑢𝑡 (𝑡):

9
HW 1 @ 2020-02-11 17:16:14-08:00

𝑑
𝐼 𝑐 𝑝𝑚𝑜𝑠 = 𝐶 𝑝𝑚𝑜𝑠 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 )
𝑑𝑡
(9)
𝑑
𝐼 𝑐 𝑛𝑚𝑜𝑠 = 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) (10)
𝑑𝑡
𝑉𝑜𝑢𝑡,1 (𝑡)
𝐼𝑅 𝑜𝑛 = (11)
𝑅 𝑜𝑛
𝐼 𝑐 𝑝𝑚𝑜𝑠 + 𝐼 𝑐 𝑛𝑚𝑜𝑠 = −𝐼𝑅 𝑜𝑛 (12)
𝑑 𝑑 𝑉𝑜𝑢𝑡,1 (𝑡)
𝐶 𝑝𝑚𝑜𝑠 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 ) + 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) = − (13)
𝑑𝑡 𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑑 𝑉𝑜𝑢𝑡,1 (𝑡)
𝐶 𝑝𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) + 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) = − (14)
𝑑𝑡 𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡)
(𝐶 𝑝𝑚𝑜𝑠 + 𝐶 𝑛𝑚𝑜𝑠 ) 𝑉𝑜𝑢𝑡,1 (𝑡) = − (15)
𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡)
𝑉𝑜𝑢𝑡,1 (𝑡) = −
𝑑𝑡 𝑅 𝑜𝑛 (𝐶 𝑝𝑚𝑜𝑠 + 𝐶 𝑛𝑚𝑜𝑠 )
(16)
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡)
𝑉𝑜𝑢𝑡,1 (𝑡) = − (17)
𝑑𝑡 2𝑅 𝑜𝑛 𝐶 𝐺

b) Given the initial conditions in part (a), solve for 𝑉𝑜𝑢𝑡,1 (𝑡).

Solution
We know that the solution to a differential equation of the form

𝑑 𝑉𝑜𝑢𝑡,1
𝑉𝑜𝑢𝑡,1 (𝑡) = −
𝑑𝑡 𝑅 𝑜𝑛 (2𝐶 𝐺 )

is
𝑡
−𝑅
𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑘𝑒 𝑜𝑛 (2𝐶 𝐺 )

Plugging in the initial condition 𝑉𝑜𝑢𝑡,1 (0) = 𝑉𝐷𝐷 we find that 𝑉𝑜𝑢𝑡,1 (𝑡) =
𝑡
−𝑅
𝑉𝐷𝐷 𝑒 𝑜𝑛 (2𝐶 𝐺 ) .

c) Sketch the output voltage of the first inverter, showing clearly (1) the
initial value, (2) the initial slope, (3) the asymptotic value, and (4) the
time that it takes for the voltage to decay to roughly 1/3 of its initial
value.

10
EECS 16B Spring 2020 HW 1

Solution

(1) We know that the output of our inverter started with the initial value
𝑉𝐷𝐷 .

(2) Since the differential equation tells us the change in value of 𝑉𝑜𝑢𝑡,1 (𝑡)
at time 𝑡 we can simply plug in 𝑡 = 0 into our differential equation to get
the initial slope:

𝑑 𝑉𝑜𝑢𝑡,1 (0)
𝑉𝑜𝑢𝑡,1 (𝑡) = − (18)
𝑑𝑡 𝑅 𝑜𝑛 (𝐶 𝑛𝑚𝑜𝑠 + 𝐶 𝑝𝑚𝑜𝑠 )
𝑑 𝑉𝐷𝐷
𝑉𝑜𝑢𝑡,1 (𝑡) = − (19)
𝑑𝑡 𝑅 𝑜𝑛 (𝐶 𝑛𝑚𝑜𝑠 + 𝐶 𝑝𝑚𝑜𝑠 )

𝑉𝐷𝐷
Thus the initial slope is − 𝑅 𝑜𝑛 (𝐶𝑛𝑚𝑜𝑠 +𝐶 𝑝𝑚𝑜𝑠 )
= − 𝑅 𝑜𝑛𝑉(𝐷𝐷
2𝐶 𝐺 )

(3) Since the input to the inverter changed from high to low we know the
output of the first inverter (𝑉𝑜𝑢𝑡,1 ) is going to go to 0 in steady state, as
this node will be discharged by the first inverter’s nmos transistor.

Alternatively, we can find the asymptotic value by plugging in 𝑡 = ∞ to



−𝑅
the solution we found for 𝑉𝑜𝑢𝑡,1 (𝑡) to find 𝑉𝑜𝑢𝑡,1 = 𝑉𝐷𝐷 𝑒 𝑜𝑛 (2𝐶 𝐺 ) = 0.

(4) To approximate when the output will decay to 13 its original value, we
use the fact that 𝑒 −1 = 1𝑒 ≈ 13 . We thus want to find when 𝑉𝑜𝑢𝑡,1 = 𝑉𝐷𝐷 𝑒 −1 .
This will occur when the 𝑒 term is raised to −1, which occurs when
𝑡 = 𝑅 𝑜𝑛 (2𝐶 𝐺 ) = 2 ∗ 10−12 .

11
HW 1 @ 2020-02-11 17:16:14-08:00

Voltage on discharging capacitor over time


1

0.8

0.6

Voltage
0.4

0.2

0
time (s)

Figure 5

d) A long time later, the input to the first inverter switches low again.
Solve for 𝑉𝑜𝑢𝑡,1 (𝑡).
Sketch the output voltage of the first inverter (𝑉𝑜𝑢𝑡,1 ), showing clearly
(1) the initial value, (2) the initial slope, and (3) the asymptotic value.

Solution

We know that after a long time, the output of the first inverter has stabilized
to 0. When the input switches low again, the input inverter’s nmos device
turns off, while the input inverter’s pmos device turns on. This connects
the 𝑉𝑜𝑢𝑡,1 node to 𝑉𝐷𝐷 , as shown in fig. 6.

12
EECS 16B Spring 2020 HW 1

𝑉𝐷𝐷 𝑉𝐷𝐷

𝑅 𝑜𝑛
𝐶 𝑝𝑚𝑜𝑠 = 𝐶

𝑉𝑜𝑢𝑡,1

𝐶 𝑛𝑚𝑜𝑠 = 𝐶

Figure 6: Inverter output at 1

To set up the differential equation, we apply KVL and KCL again:

𝑑
𝐼 𝑐 𝑝𝑚𝑜𝑠 = 𝐶 𝑝𝑚𝑜𝑠 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 )
𝑑𝑡
(20)
𝑑
𝐼 𝑐 𝑛𝑚𝑜𝑠 = 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) (21)
𝑑𝑡
𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
𝐼𝑅 𝑜𝑛 = (22)
𝑅 𝑜𝑛
𝐼 𝑐 𝑝𝑚𝑜𝑠 + 𝐼 𝑐 𝑛𝑚𝑜𝑠 = −𝐼𝑅 𝑜𝑛 (23)
𝑑 𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
𝐶 𝑝𝑚𝑜𝑠 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 ) + 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) = − (24)
𝑑𝑡 𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
𝐶 𝑝𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) + 𝐶 𝑛𝑚𝑜𝑠 𝑉𝑜𝑢𝑡,1 (𝑡) = − (25)
𝑑𝑡 𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
(𝐶 𝑝𝑚𝑜𝑠 + 𝐶 𝑛𝑚𝑜𝑠 ) 𝑉𝑜𝑢𝑡,1 (𝑡) = − (26)
𝑑𝑡 𝑅 𝑜𝑛
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
𝑉𝑜𝑢𝑡,1 (𝑡) = −
𝑑𝑡 𝑅 𝑜𝑛 (𝐶 𝑝𝑚𝑜𝑠 + 𝐶 𝑛𝑚𝑜𝑠 )
(27)
𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷
𝑉𝑜𝑢𝑡,1 (𝑡) = − (28)
𝑑𝑡 2𝑅 𝑜𝑛 𝐶 𝐺

13
HW 1 @ 2020-02-11 17:16:14-08:00

We will use substitution of variables:


𝑥(𝑡) = 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 (29)
𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑥(𝑡) + 𝑉𝐷𝐷 (30)
𝑑 𝑑
𝑥(𝑡) = 𝑉𝑜𝑢𝑡,1 (𝑡) (31)
𝑑𝑡 𝑑𝑡
Substituting in:
𝑑 𝑥
𝑥(𝑡) = − (32)
𝑑𝑡 2𝑅 𝑜𝑛 𝐶 𝐺
− 2𝑅 𝑜𝑛𝑡 𝐶
𝑥(𝑡) = 𝐴𝑒 𝐺 (33)

Substituting again for 𝑥(𝑡):


− 2𝑅 𝑜𝑛𝑡 𝐶
𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑉𝐷𝐷 + 𝐴𝑒 𝐺

Using the initial condition 𝑉𝑜𝑢𝑡,1 = 0 (as the input to the first inverter was
high for a long time before switching low) implies 𝐴 = −𝑉𝐷𝐷 . Thus:
− 2𝑅 𝑜𝑛𝑡 𝐶
 
𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑉𝐷𝐷 1 − 𝑒 𝐺

(1) Because the input to the first inverter was high for a long time, we
know the initial value of 𝑉𝑜𝑢𝑡,1 (𝑡) = 0. This was the initial condition
applied to the solution of the differential equation, above.

(2) To find the initial value of the slope we can plug in 𝑡 = 0 to the
above differential equation:
𝑑 (𝑉𝐷𝐷 − 𝑉𝑜𝑢𝑡,1 (0))
𝑉𝑜𝑢𝑡,1 (𝑡) =
𝑑𝑡 𝑅 𝑜𝑛 (2𝐶 𝐺 )
(𝑉 )
where 𝑉𝑜𝑢𝑡,1 (0) = 0. Thus our initial slope is 𝑅 𝑜𝑛 (𝐷𝐷
2𝐶 𝐺 )
. Notice this slope is
positive while the previous part had a negative slope.
(3) Since the input to the inverter changed from low to high and the input
inverter’s pmos is now on, we know the output of the first inverter is
going to go to 𝑉𝐷𝐷 in steady state.

Alternatively, we can find the asymptotic value by plugging in 𝑡 = ∞ to


− 𝑅 (∞2𝐶 )

the solution we found for 𝑉𝑜𝑢𝑡,1 (𝑡) to find 𝑉𝑜𝑢𝑡,1 = 𝑉𝐷𝐷 1 − 𝑒 𝑜𝑛 𝐺 =
𝑉𝐷𝐷 (1 − 0) = 𝑉𝐷𝐷 .

14
EECS 16B Spring 2020 HW 1

Voltage on charging capacitor over time


1

0.8

0.6
Voltage

0.4

0.2

0
time (s)

Figure 7

e) For each complete input cycle described above (𝑉𝑖𝑛 = 0V → 1V → 0V),


how much charge is pulled out of the power supply? Give both a
symbolic and numerical answer. Consider only the charge needed to
charge up the 𝑉𝑜𝑢𝑡,1 node.

Solution
To find the charge required from the supply, we can integrate ∫the current re-

quired from the supply during each phase of the cycle ( 𝑄 = 0 𝐼𝑉𝐷𝐷 (𝑡)𝑑𝑡).
During the input step from 𝑉𝑖𝑛 = 0 to 𝑉𝑖𝑛 = 𝑉𝐷𝐷 , we know that the
𝑡
− 2𝐶
voltage is 𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑉𝐷𝐷 𝑒 𝐺 𝑅 𝑜𝑛 . We then get:

𝑑
𝐼𝑉𝐷𝐷 = 𝐼𝐶 𝑝𝑚𝑜𝑠 = 𝐶 𝐺 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 )
𝑑𝑡
−1 − 𝑡
= 𝐶𝐺 𝑉𝐷𝐷 𝑒 2𝐶𝐺 𝑅 𝑜𝑛
2𝐶 𝐺 𝑅 𝑜𝑛

Thus:
∫ ∞ ∫ ∞
−1 − 𝑡
𝑄 0→1 = 𝐼𝐶 𝑝𝑚𝑜𝑠 (𝑡)𝑑𝑡 = 𝐶𝐺 𝑉𝐷𝐷 𝑒 2𝐶𝐺 𝑅 𝑜𝑛 𝑑𝑡
0 0 2𝐶 𝐺 𝑅 𝑜𝑛
𝑡
− 2𝐶
= 𝐶 𝐺 · 𝑉𝐷𝐷 𝑒 𝐺 𝑅 𝑜𝑛 |0∞
= 𝐶 𝐺 · 𝑉𝐷𝐷 (0 − 1) = −𝐶 𝐺 · 𝑉𝐷𝐷

15
HW 1 @ 2020-02-11 17:16:14-08:00

During the input step from  𝑉𝑖𝑛 = 𝑉 𝐷𝐷 to 𝑉𝑖𝑛 = 0, we know that the
− 2𝑅 𝑜𝑛𝑡 𝐶
voltage is 𝑉𝑜𝑢𝑡,1 (𝑡) = 𝑉𝐷𝐷 1 − 𝑒 𝐺 . The current from the supply

will be equal to the sum of the resistor current and PMOS gate capacitor
current: 𝐼𝑉𝐷𝐷 = 𝐼𝑅 + 𝐼𝐶 𝑝𝑚𝑜𝑠 . We get:

𝑑 𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷


𝐼𝑉𝐷𝐷 = 𝐼𝐶 𝑝𝑚𝑜𝑠 + 𝐼𝑅 = 𝐶 𝐺 (𝑉𝑜𝑢𝑡,1 (𝑡) − 𝑉𝐷𝐷 ) +
𝑑𝑡 𝑅 𝑜𝑛
− 2𝑅 𝑜𝑛𝑡 𝐶
 
−1 𝑉𝐷𝐷 1 − 𝑒 𝐺 − 𝑉𝐷𝐷
− 2𝐶 𝑡𝑅 𝑜𝑛
= −𝐶 𝐺 𝑉𝐷𝐷 𝑒 𝐺 +
2𝐶 𝐺 𝑅 𝑜𝑛 𝑅 𝑜𝑛
− 2𝑅 𝑜𝑛𝑡 𝐶
−1 − 𝑡 −𝑉𝐷𝐷 𝑒 𝐺
= −𝐶 𝐺 𝑉𝐷𝐷 𝑒 2𝐶𝐺 𝑅 𝑜𝑛 +
2𝐶 𝐺 𝑅 𝑜𝑛 𝑅 𝑜𝑛

Thus:
𝑡
∞ ∞ −
−𝑉𝐷𝐷 𝑒 2𝑅 𝑜𝑛 𝐶𝐺
∫ ∫
−1 − 𝑡
𝑄 1→0 = 𝐼𝐶 𝑝𝑚𝑜𝑠 (𝑡) + 𝐼𝑅 (𝑡)𝑑𝑡 = −𝐶 𝐺 𝑉𝐷𝐷 𝑒 2𝐶𝐺 𝑅 𝑜𝑛 + 𝑑𝑡
0 0 2𝐶 𝐺 𝑅 𝑜𝑛 𝑅 𝑜𝑛
− 2𝐶 −𝑉𝐷𝐷
𝑡
− 𝑡
= −𝐶 𝐺 𝑉𝐷𝐷 𝑒 |0∞ +𝐺 𝑅 𝑜𝑛 · −1 · 2𝑅 𝑜𝑛 𝐶 𝐺 𝑒 2𝑅 𝑜𝑛 𝐶𝐺 |0∞
𝑅 𝑜𝑛
= −𝐶 𝐺 𝑉𝐷𝐷 (0 − 1) + 2𝐶 𝐺 𝑉𝐷𝐷 (0 − 1)
= −𝐶 𝐺 𝑉𝐷𝐷

The total charge is thus

𝑄 𝑡𝑜𝑡𝑎𝑙 = 𝑄 0→1 + 𝑄1→0 = −𝐶 𝐺 𝑉𝐷𝐷 − 𝐶 𝐺 𝑉𝐷𝐷 = −2𝐶 𝐺 𝑉𝐷𝐷

Note that the current direction for 𝐼𝑉𝐷𝐷 was pointing into the 𝑉𝐷𝐷 source,
so the charge represents the charge moved into the power supply. As the
question asks for the charge pulled out of the power supply, we know:

𝑄 pulled out of power supply = −𝑄 𝑡𝑜𝑡𝑎𝑙 = 2𝐶 𝐺 𝑉𝐷𝐷 = 2(1fF · 1V) = 2fC

Alternative solution:
During the input step when 𝑉𝑖𝑛 = 0, note that 𝑉𝑜𝑢𝑡,1 is connected to 𝑉𝐷𝐷
through the input inverter’s pmos. Thus, during this phase, the power
supply is supplying charge to change the node voltage at node 𝑉𝑜𝑢𝑡,1 . We
use the equation
𝑄 = 𝐶𝑉

16
EECS 16B Spring 2020 HW 1

noting that 𝑉𝑖𝑛𝑖𝑡𝑖𝑎𝑙 = 0 and 𝑉 𝑓 𝑖𝑛𝑎𝑙 = 𝑉𝐷𝐷 .


Thus:

𝑄 = 𝐶𝑉
= (𝐶 𝑛𝑚𝑜𝑠 + 𝐶 𝑝𝑚𝑜𝑠 )(𝑉𝐷𝐷 − 0)
= 2𝐶 𝐺 𝑉𝐷𝐷
= 2(1fF · 1V) = 2fC

During the input step when 𝑉𝑖𝑛 = 1, note that 𝑉𝑜𝑢𝑡,1 is connected to
ground through the input inverter’s nmos. In this case, the supply is not
providing any charge to the 𝑉𝑜𝑢𝑡,1 node. Rather, the charge on this node
is being moved to ground through the nmos. Thus there is no charge
during this input step.

For the entire input cycle, we thus find that 𝑄 = 2𝐶 𝐺 𝑉𝐷𝐷 = 2fC.

4 RC Circuit
Consider the circuit below, assume that when 𝑡 ≤ 0, the capacitor has no charge
stored (𝑉c (𝑡 = 0) = 0). At 𝑡 = 0, the switch closes. Assume that 𝑉𝑠 = 5 V,
𝑅 = 100 Ω, and 𝐶 = 10 µF.

𝑡=0 𝑅
𝐼𝑅 (𝑡)
+ −
𝑉𝑅 (𝑡)
+ 𝐼 𝑐 (𝑡)

+
𝑉𝑠 𝑉𝑐 (𝑡) 𝐶

Figure 8: RC Circuit with Voltage Source

a) What are the boundary conditions for 𝐼c (𝑡) (i.e. what are 𝐼c (𝑡 = 0) and
𝐼c (𝑡 → ∞)?)

Solution

No charge is on the capacitor before time 𝑡 = 0. Using 𝑞 = 𝑉 𝐶, we know


that 𝑉c = 0 V before 𝑡 = 0.

17
HW 1 @ 2020-02-11 17:16:14-08:00

At 𝑡 = 0, the switch closes, so we can use KVL to say

𝑉c + 𝑉R − 𝑉s = 0

Since voltage across a capacitor cannot change instantaneously,

𝑉c (𝑡 = 0) = 0

which means

𝑉R (𝑡 = 0) = 𝑉s

From here we can use Ohm’s Law to get

𝑉s 5
𝐼c (𝑡 = 0) = = = 0.05 A
𝑅 100
As 𝑡 goes to infinity, the capacitor will become fully charged and block all
current. Therefore

𝐼c (𝑡 → ∞) = 0 A


b) Use KVL and the relationship between charge and current (𝑞 = 𝐼𝑑𝑡) to
find the first order differential equation in terms of the current through
the capacitor, 𝐼c . Assume that 𝑑𝑉 s
𝑑𝑡 = 0. (Hint: You will need to take a
derivative with respect to time to get the equation.)

Solution

From KVL we get:

𝑉1 + 𝑉R − 𝑉s = 0

Using Ohm’s law and 𝑞 = 𝑉 𝐶:


𝑞
+ 𝐼𝑅 𝑅 − 𝑉s = 0
𝐶
Since 𝑅 and 𝐶 are in series, we can say:

𝐼 𝑅 = 𝐼c

Substituting 𝑞 in terms of 𝐼, we get:



𝐼c 𝑑𝑡
+ 𝐼c 𝑅 − 𝑉s = 0
𝐶

18
EECS 16B Spring 2020 HW 1

𝑑𝐼c
To get the differential equation in terms of 𝐼c and 𝑑𝑡 , we take the derivative
with respect to time:

𝐼c 𝑑𝐼c 𝑑𝑉s
+𝑅 − =0
𝐶 𝑑𝑡 𝑑𝑡
𝑑𝑉s
𝑑𝑡 = 0, so we end up with:

𝐼c 𝑑𝐼c
+𝑅 =0
𝐶 𝑑𝑡

c) What is the eigenvalue 𝜆 of this equation?

Solution

We can rearrange the equation to

𝑑𝐼c 𝐼c
=−
𝑑𝑡 𝑅𝐶
Since this is a first order differential equation, 𝜆 is equal to the coefficient
of the 𝐼c term.
1
𝜆=−
𝑅𝐶

d) Using the eigenvalue and boundary conditions found in the previous


parts, find an expression for 𝐼c (𝑡) in terms of 𝑉s , 𝑅, and 𝐶.

Solution

The general solution to the equation

𝑑𝑦
= 𝜆𝑦
𝑑𝑡
is

𝑦(𝑡) = 𝐾𝑒 𝜆𝑡

where 𝐾 is a constant and 𝜆 is the eigenvalue of the equation. To find 𝐾,


we can plug in our boundary condition at 𝑡 = 0

0 𝑉s
𝐼c (𝑡 = 0) = 𝐾𝑒 − 𝑅𝐶 =
𝑅

19
HW 1 @ 2020-02-11 17:16:14-08:00

From this we can see


𝑉s
𝐾=
𝑅
So our overall equation ends up being
𝑉s − 𝑡
𝐼c (𝑡) = 𝑒 𝑅𝐶
𝑅
We can also double check our answer by looking at the steady state
boundary condition:
𝑉s −∞
𝐼c (𝑡 → ∞) = 𝑒 =0
𝑅
which agrees with our answer from part (a).

e) On what order of magnitude of time (nanoseconds, milliseconds, 10’s of


seconds, etc.) does this circuit settle (𝐼c is < 5% of its initial value)?

Solution

The time constant 𝜏 of an RC circuit is just 𝜏 = 𝑅𝐶. For our circuit:


𝜏 = 𝑅𝐶 = 100 Ω · 10 µF = 0.001 s
After 3 time constants, the current will be 5% of its initial value
3𝜏 = 0.003 s
The circuit will settle on the order of milliseconds.

f) Give 2 ways to reduce the settling time of the circuit if we are allowed to
change one component in the circuit.

Solution
To reduce settling time, reduce 𝜏. We can achieve this by

a) Lowering the value of 𝑅 or


b) Lowering the value of 𝐶.

Notice how the value of 𝑉s does not change the settling time.
g) Sketch the current vs. time plot of 𝐼c (𝑡). Make sure to label 𝐼c (𝑡) at 𝑡 = 0,
𝑡 = 𝜏, 𝑡 = 2𝜏, and 𝑡 = 3𝜏.

20
EECS 16B Spring 2020 HW 1

Solution

𝑉s − 𝑡
𝐼c (𝑡) = 𝑒 𝑅𝐶
𝑅
𝐼c (𝑡) = 0.05𝑒 −1000𝑡
𝐼c (0) = 0.05 A
𝐼c (𝜏) = 18.3 mA
𝐼c (2𝜏) = 6.76 mA
𝐼c (3𝜏) = 2.48 mA

50

40

30
𝐼c [mA]

20

10

0 1 2 3 4 5
𝑡[ms]

5 (OPTIONAL) Write Your Own Question And Provide a


Thorough Solution.
Writing your own problems is a very effective way to really learn material.
Having some practice at trying to create problems helps you study for exams
much better than simply solving existing practice problems. This is because
thinking about how to create an interesting problem forces you to really
consolidate your understanding of the course material.

6 Homework Process and Study Group


Citing sources and collaborators are an important part of life, including being
a student! We also want to understand what resources you find helpful and
how much time homework is taking, so we can change things in the future if
possible.

21
HW 1 @ 2020-02-11 17:16:14-08:00

a) What sources (if any) did you use as you worked through the home-
work?
b) If you worked with someone on this homework, who did you work
with? List names and student ID’s. (In case of homework party, you can
also just describe the group.)
c) How did you work on this homework? (For example, I first worked by
myself for 2 hours, but got stuck on problem 3, so I went to office hours. Then I
went to homework party for a few hours, where I finished the homework.)
d) Roughly how many total hours did you work on this homework?

22

You might also like