Mixed signal circuit design -DAC architecture Part1
Oct 6, 2023
Now let's see the DAC architecture
We have discussed DAC specification and ADC specification
Now let's see the DAC architecture
There are variety of DAC architecture
From simple architecture to complex architecture
Each architecture has its own merits
Now let's see the digital input code used in DAC
In DAC, the input signal is digital
Let's see the digital input code
The digital input code is binary
Binary code is the digital input code for 1 and 0
It doesn't have to be a binary number
It can be binary, bcd, thermometer, gray, sine, two's complement, offset, etc.
The advantage of the digital input code is
It can be changed from one bit to another
That's why we use other codes
For example, in thermometer code, changes are made for each bit
0 is changed to 1
Transition from 1 to 0 is changed to 1
In gray code, 0 is changed to 1
0 is changed to 1
The reason why we use binary code is
It is desirable to change from one bit to another
These are the different digital input codes we use
Bcd, thermometer, gray, sine, two's complement, offset, etc.
The first type of DAC is Resistor String DAC
Let's see the Resistor String DAC
We use identical resistors in Resistor String DAC
Look at the first figure
R1, R2, R3, R4 are all identical resistors
We use identical resistors in R1, R2, R3, R4
We also use switches in R1, R2, R4
We use switches in R1, R2, R4
We use an N is to 2 raised to N decoder to control 2 raised to N signals
Accuracy of Resistor String DAC is very high
Resistor String DAC output is always monotonic
The problem with Resistor String DAC is
The output is always connected to 2 raised to N switches
When connected to 2 raised to N switches,
If we use larger resolution,
Large parasitic capacities will be affected
The speed of the output load will decrease
The problem with Resistor String DAC is
The output is always connected to 2 raised to N switches
If we use larger resolution,
Large parasitic capacities will be affected
The speed of the output load will decrease
Resistor String DAC is a better alternative to Figure-B
Resistor String DAC is a better alternative to Figure-A
Binary switch array ensures that output is connected to N switches that are on and N switches that are off
Here we are using the binary switch array.
In this, we are working with N switches in ON condition and N switches in OFF condition.
So we can improve the conversion speed in this case.
The problem of resistor string DAC is the balance between area and power dissipation.
If the chip area is high bit resolution, we need more parasitic components.
So we need more resistors for high bit resolution.
So the size of the chip is more and power dissipation is more.
This is the main disadvantage of resistor string DAC.
Now let's see the mismatching error related to resistor string DAC.
The accuracy of resistor string is obviously related to matching between resistors which ultimately determines INL and DNL for the entire DAC.
The accuracy of a resistor string DAC is always related to matching between all resistors.
The matching between INL and DNL depends on this.
Now suppose that the ith resistor Ri has a mismatch error associated with it.
I am assuming an ith resistor Ri.
Let's assume that this ith resistor has a mismatch error.
Then the ith resistance value of ith resistor is Ri is equal to R plus delta Ri.
R is the ideal resistance value.
Delta Ri is the mismatch error.
Then the value of ith resistance is Ri is equal to R plus delta Ri.
Now let's assume that the mismatch error is symmetrical.
That is, if the mismatch error is same everywhere, then the sum of all mismatch terms is zero.
Then we can write I is equal to 1 to 2 rise to n delta Ri is equal to zero.
The value of voltage at the tab associated with ith resistor should ideally be.
This is an equation to find the value of the voltage coming across the ith resistor.
That is, each resistor in this resistor string works like a voltage divider circuit.
The equivalent analog voltage value is given in the output.
The ideal value is I into Vref divided by 2 rise to n.
That is, the value of the voltage coming across the resistance I is I into Vref divided by 2 rise to n.
The value of I can be 0, 1, 2, etc. up to 2 rise to n.
However, including the mismatch, the actual value of ith voltage is the sum of the resistance up to including resistance I divided by sum of all the resistance in the string.
The value of I is the value in the ideal case.
The equation to find the value of Vi is Vref into sigma k is equal to 1 to I Rk divided by sigma k is equal to 1 to 2 rise to n Rk.
The actual value of ith voltage is the sum of all resistance value up to and including resistance I divided by sum of all resistance in the string.
That is, sigma k is equal to 1 to 2 rise to n Rk.
The equation to find the value of V2 is Vref into sigma k is equal to 0 to 2 rise to n Rk.
That is, sigma k is equal to 0 to 2 rise to n Rk divided by sum of all resistance value up to and including resistance I divided by sum of all resistance in the string.
This is how we find the value of ith voltage.
Thus, we get the corresponding output value for V1, V2, V3, etc.
Each resistor acts as a voltage divider and we get the corresponding output value for each switch.
This is how a resistor string works.
We get Vi is equal to Vref into sigma k is equal to 1 to I Rk divided by sigma k is equal to 1 to 2 rise to n Rk.
We can modify this equation in the same way.
Vi is equal to Vref into sigma k is equal to 1 to I Rk divided by sigma k is equal to 1 to I Rk.
Similarly, Rk is equal to R plus delta Rk.
R is equal to R plus delta Rk divided by sigma k is equal to 1 to 2 rise to n Rk.
Now, let's see what 1 to 2 rise to n is.
Rk is equal to R1 plus R2 plus R3 plus etc. up to and including R2 rise to n.
If all the values of R are identical, we can write 2 rise to n into Rn.
For example, if sigma k is equal to 1 to 3 Rk, we can write R1 plus R2 plus R3.
If all the values of R1, R2 and R3 are identical, we can write R plus R plus R is equal to 3R.
Similarly, we can write 1 to 2 rise to n Rk is equal to 2 rise to n into R.
The denominator does not include any mismatch error.
Since it is assumed that the mismatch sum is zero,
we can write sigma k is equal to 1 to i R is equal to R plus R plus R is equal to 2 rise to n Rk.
Similarly, we can write this equation as Vref divided by 2 rise to n R into sigma k is equal to 1 to i R is equal to R plus R plus R plus etc.
Instead of writing R to i, we can write it as i into R plus sigma k is equal to 1 to i delta Rk.
Similarly, we can write Vref divided by 2 rise to n R into i into R is equal to 1 to n delta Rk divided by R.
Finally, we got the equation Vi is equal to Vi ideal.
This is the equation we studied earlier.
We got the equation for Vi ideal as i into Vref divided by 2 rise to n.
We got the equation for Vi ideal as i into Vref divided by 2 rise to n into sigma k is equal to 1 to n delta Rk divided by R.
Now, we find the nonlinearity in the resistor string.
Integral nonlinearity is defined as the difference between actual and ideal switching points.
The equation for integral nonlinearity is Vi minus Vi ideal.
We got the value of Vi here.
This is the value of Vi ideal.
When we subtract these two, we get Vi minus Vi ideal.
We can also take Vi ideal here.
We got the equation for integral nonlinearity as Vref divided by 2 rise to n into sigma k is equal to 1 to n delta Rk divided by R.
Equation 1 is the general equation for integral nonlinearity for a given resistor Ri.
Hope you understood.

