Mixed Signal Circuit Design -DAC Specification
Oct 6, 2023
Module 6
This figure is a block diagram.
N-bit word is mapped into a single analog voltage.
Here, we are giving N-bit word as digital word.
Digital to analog converter uses it as a single analog output.
Here, there is a reference voltage.
Here, we get a fraction of the reference voltage.
Output is a fraction of reference voltage.
So, Vout is equal to F into Vreference.
Vout is the analog output.
Vreference is the reference voltage.
F is the fraction.
Fraction is defined by input word d.
Fraction is the fraction of the reference voltage.
Number of input combination is 2 raised to N.
Let's see what fraction is.
If the input is N-bit word,
then the value of fraction can be determined by d divided by 2 raised to N.
Fraction is equal to d divided by 2 raised to N.
Example
A 3-bit DAC is being used.
The input d is 1 0 0.
Decimal value of 1 0 0 is 4.
If the reference voltage is 5 V,
then fraction is equal to d divided by 2 raised to N.
d is the corresponding decimal value.
4 divided by 2 raised to N.
How many digits are there?
1, 2, 3.
So, 2 cube.
Fraction is 1 by 2.
Analog voltage that appears at the output becomes
Vout is equal to F into Vreference.
The same equation is given here.
Vout is equal to F into Vreference.
Fraction is 1 by 2 into 5.
That is equal to 2.5 V.
This is the output voltage.
Fraction of input voltage in the case of digital to analog converters.
Next is the full-scale voltage.
The maximum output voltage can be generated is known as full-scale voltage.
Vout is equal to 2 raised to N minus 1 divided by 2 raised to N into Vreference.
This is the maximum output voltage that can be generated by a DAC.
Next is the load voltage.
Load voltage is the maximum output voltage that can be generated by a DAC.
This is the equation of full-scale voltage.
Next is the least significant bit and the most significant bit.
Let's see what the least significant bit is.
The least significant bit is the rightmost bit of a digital wave.
For example, if we write 1 0 0, this 0 is the least significant bit.
Let's see how much 1 LSB affects the analog output.
1 LSB is equal to Vreference divided by 2 raised to N.
Vreference divided by 2 raised to N is based on this equation.
The most significant bit or MSB refers to the leftmost bit of the digital wave D.
MSB is the leftmost bit of the digital wave D.
MSB is equal to Vreference divided by 2 raised to N.
LSB is equal to Vreference divided by 2 raised to N.
Next is the resolution of a DAC.
The smallest change in analog output with respect to the value of reference voltage Vreference is called resolution.
This is slightly different from the definition of LSB.
In that, resolution is typically given in terms of bits and represents the number of unique output voltage levels.
Next, let's look at differential nonlinearity.
Next, let's look at differential nonlinearity or DNL.
DNL is an ideal transfer curve of a 3-bit DAC.
The y-axis is given as a normalized value.
The difference between each axis is 1 by 8.
That is, everything from here to here is the same value.
Everything adjacent to it is going as a unit.
To convert this to voltage, we multiply the adjacent heights by Vreference.
Increment heights can be converted to volts by multiplying height by Vreference.
Multiplying height by Vreference will convert it to equivalent analog value.
That is, 1 LSB is equal to 1 by 8.
Next is the difference of 1 by 8.
Assuming the Vreference value is 5 volts,
to get the output voltage, we multiply this by 5 volts.
That is, 1 by 8 into LSB is equal to 1 by 8 into 5 volts.
That is 5 by 8.
5 by 8 is equal to 0.625.
There is a zero in it. That is a mistake.
0.625 is the value of 1 LSB.
Non-ideal components cause analog increment to differ from the ideal value.
The difference between ideal and non-ideal value is called differential nonlinearity.
Let us write the equation in DNL.
Actual Incremental Height of Transmission n minus Ideal Increment Height.
The difference between Actual Increment Height and Ideal Increment Height is called differential nonlinearity.
Where n is corresponding to the digital input transmission.
We have given an example of this differential nonlinearity error.
An example 4 of differential nonlinearity for a 3-bit DAC.
We have split it into 3 bits.
This is the corresponding digital value.
0 0 0 is equal to 0.
0 0 1 is equal to 1 by 8.
1 by 8 into Vref is equal to analog value Vout.
This is Vout divided by Vref.
This is how we drew the first point and the second point.
This is the ideal slope.
This is the ideal height from 0 to 1 by 8.
This is the ideal height from 1 by 8 to 2 by 8.
This is the ideal height.
This is a constant value.
Differential nonlinearity is the variation from the ideal height.
This is the ideal height.
But the next step is not up.
The ideal height is 1.5 into the ideal height.
The next value is 1.
But in this case, the ideal height is 0.5.
The ideal height is 0.5.
The next value is 0.25.
The next value is 1.75.
This is an example figure for differential nonlinearity.
We can find out the differential nonlinearity error in each case.
This is the differential nonlinearity error 3.
We can find out the differential nonlinearity error 3.
We need to find out the unit of the point 3.
What is the change in that point?
We can find out the actual increment height of transmission in differential nonlinearity error 1.5.
The ideal height is 1.
1.5 LSB minus 1.5 LSB is equal to 0.5 LSB.
Now, look at the difference in the fourth point.
The height from ideal height to 4 is 0.5
Then we can write 0.5-1
0.5 lsb-1 lsb is equal to 0.5 lsb
Similarly, dnl4 is equal to 0.5 lsb-1 lsb is equal to minus 0.5 lsb
Similarly, we can find dnl5
dnl5 is equal to 0.25 lsb-1 lsb is equal to 0.75 lsb
Similarly, we can find dnl6
dnl6 is equal to 1.75-1 lsb-1 lsb is equal to 0.75 lsb
In the case of dnl7, the ideal height itself came
So, dnl error is 0
In this way, we find dnl error
This is a dnl curve for non-ideal 3d 3-bit DAG
This is the dnl curve we drew earlier
Here, take dnl lsb in the y-axis
In the x-axis, we give the digital input code
When we come to this figure, we notice that there is no error here and here
Here we got the first error 0.5 lsb
Next, we got 0.5 lsb
Next, we got 0.75 lsb
Next, we got 0 lsb
That's what we plotted here
There is no error in 0 and 0.1
There is no error in 2
When we came to 3, we got 0.5 lsb
When we came to 4, we got 0.5 lsb
When we came to 5, we got 0.75 lsb
When we came to 6, we got 0.75 lsb
When we came to 7, we got 0 lsb
This is the dnl curve we studied earlier
The next effect is called Integral Nonlinearity or INL
INL stands for Integral Nonlinearity Error
This figure has two end points
We are drawing a straight line based on these two end points
The value of the dac from the straight line is how much it is changing
This is called Integral Nonlinearity Error
All other arrays due to offset and gain are 0.5 lsb
Assume that the converter with n-bit resolution will have less than ±2 1x2 lsb of dnl and inl
The term 1x2 lsb is a common term that typically denotes the maximum error of a data converter, both DAC and ADC
For example, a 13-bit DAC having greater than ½ lsb of dnl or inl actually has a resolution of 12-bit DAC
Therefore, the value of 1x2 lsb should be less than ½ lsb of dnl or inl
The value of ½ lsb in volts is 0.5 or ½ lsb is equal to Vref divided by 2 raised to n plus 1
There are other methods to determine INL error
Compare the output value to ideal reference line regardless of positions of first and last output values
But here we will consider offset voltage and gain error as an extra
Another method is INL determination
Another method described as best fit method
Attempts to minimize INL error constructing reference line so that it passes closely as possible to majority of output
Now, let's look at offset
The analog output should be 0 for D is equal to 0
So, what will happen to the transfer carrier if the offset voltage is 0?
There will be a change
That's what is shown in this figure
X and Y are the same as before
But here it should have started from 0
Because here an offset voltage exists, instead of starting from 0, it started from 1 by 8
The reason why the transfer carrier shifted is because of the presence of offset
Now, let's look at gain error
Gain error exists if the slope of best fit line through the transfer curve is different from the slope of best fit line for the ideal case
Gain error is the difference between the ideal case and the transfer curve
Gain error is the equation of ideal slope minus actual slope
Now, let's look at latency
Latency is the time delay that is taken to convert the digital input to analog output
This specification defines the total time from the moment that input digital word changes to time the analog output value are settled to within specified tolerance
Latency includes delay required to map the digital word to an analog value plus the settling time
Signal to noise ratio
Signal to noise ratio is defined as the ratio of signal power to noise power
Dynamic range
Dynamic range is the ratio of largest output signal over the smallest output signal
Dynamic range is related to the resolution of the converter
An n-bit DAC can produce a maximum output of 2 raised to n-1 multiples
If the minimum value is 1 LSB, the dynamic range is maximum divided by minimum
To convert it to decibel, we have given 20 log
We get 6.02
These are the specifications of DAC
Study it carefully
Thank you

