Gain Error and Monte Carlo Analysis

Using SPICE Monte Carlo Tool
for Statistical Error Analysis
TIPL 4203
TI Precision Labs – ADCs
Created by Art Kay
Presented by Peggy Liska
1
Discrete resistor tolerance sets gain error
𝐺𝑑𝑖𝑓𝑓
𝑅𝑓
=2βˆ™
𝑅𝑔
𝐺𝑑𝑖𝑓𝑓
2π‘˜Ξ©
=2βˆ™
= 0.04
100π‘˜Ξ©
Vneg = - Vinβˆ™(Rf/Rg)+Vcm
Vneg = -(±100V)βˆ™(2k/100k)+2.5V
Vneg = 4.5V to 0.5V
Rg1 100k
Rf1 2k
5V
3V
3V
Vref
AVDD
DVDD
Vin
±100
π‘‰π‘œπ‘’π‘‘ = 𝐺𝑑𝑖𝑓𝑓 βˆ™ 𝑉𝑖𝑛
Rg2 100k
π‘‰π‘œπ‘’π‘‘ = 0.04 βˆ™ ±100𝑉 = ±4𝑉
VCM
2.5V
1% 100kΩ
Rf2 2k
U1
+
Vdif = ±4V
Rf3 2k
ADS9110
+
Rg3 100k
Min
Limit
-3Οƒ
-3Οƒ
99k
-
Max Limit
+3Οƒ
-2Οƒ
-1Οƒ
0
100k
1Οƒ
2Οƒ
3Οƒ
101k
Devices outside the minimum and maximum
limits are discarded
Rg4 100k
U2
AGND
DGND
Rf4 2k
Vpos = Vinβˆ™(Rf/Rg)+Vcm
Vneg = (±100V)βˆ™(2k/100k)+2.5V
Vneg = 0.5V to 4.5V
2
DC Transfer Function
100k
2k
5V
3V
3V
Vref
AVDD
DVDD
Vin
±100
100k
VCM
2.5V
1k
U1
+
Vdif = ±4V
ADS9110
1k
+
100k
-
100k
T
U2
AGND
DGND
2k
4.0
Vdif
0.0
-4.0
4.5
Select β€œVin” for the
input and enter the
range from β€œStart
value” to β€œEnd value”
Vneg
2.5
500.0m
4.5
Vpos
2.5
500.0m
-100
-50
0
Input voltage (V)
50
100
3
Set tolerance on resistors and capacitors
Set to 1% or 0.1% per
resistor spec
Click here to set the
tolerance.
Gaussian Distribution
4
Monte Carlo Analysis
Select β€œMonte Carlo”
99.73% sets the
component tolerance to
±3 Standard deviations
Note the default of 68.26% will
not give realistic results for
resistor and capacitor tolerance.
More cases will give you a better statistical
distribution. The max is 1000.
5
Monte Carlo for DC Transfer Characteristic
2. Press CTRL+Alt
to select all
3. Select Process>Statistics
1. Run a dc transfer
characteristic
6
The cut option
T -3.50
Zoom in
-3.65
Voltage (V)
We generate the
distribution from the
vertical section at x=100
-3.80
-3.95
-4.10
90.00
95.00
Input voltage (V)
100.00
Use Cut
option
Enter the X-axis value
where the cut set is taken
Number of bins in histogram.
10 is usually sufficient
7
Generate the statistical data and histogram
Samples
T
Press Calculate to get
statistics, and draw to
show histogram
219
197
175
153
131
109
88
66
44
22
0
-4.03
Use to calculate gain error.
-4.02
-4.01
-4.00
Values
-3.99
-3.98
-3.97
Press draw to get a graph
of the histogram
π‘ π‘‘π‘Žπ‘›π‘‘π‘Žπ‘Ÿπ‘‘ π‘‘π‘’π‘£π‘–π‘Žπ‘‘π‘–π‘œπ‘›
9.7π‘š
π‘‡π‘¦π‘πΊπ‘Žπ‘–π‘›πΈπ‘Ÿπ‘Ÿπ‘œπ‘Ÿ =
βˆ™ 100 =
βˆ™ 100 = ±0.24%
π‘€π‘’π‘Žπ‘›
4
For 68.26% of the population
π‘€π‘Žπ‘₯πΊπ‘Žπ‘–π‘›πΈπ‘Ÿπ‘Ÿπ‘œπ‘Ÿ = 3 βˆ™ π‘‡π‘¦π‘π‘–π‘π‘Žπ‘™ = 3 βˆ™ ±0.24% = 0.73%
For 99.73% of the population
8
Thanks for your time!
Please try the quiz.
9