Transcript No Slide Title

1
Motivation
Testability is an important quality metric in electronic
systems
Cost of test is beginning to dominate the per-part cost of
mant silicon products, e.g., in consumer market
Time-to-market pressures force the need of structured,
repeatable and automatic test features as part of overall
design methodology
In system-on-chip designs, test data management and
cost of test are important
Reusable IP blocks need designed-in test features
Learn terminology
Industrial demand, employment opportunities
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
2
Testing and Diagnosis
Testing: the system is exercised and the response is
analyzed to check whether it behaved correctly
Diagnosis: if incorrect behavior is detected, locate the
cause of misbehavior.
- Assumes knowledge of the internal structure of the
system
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
3
Main Categories of Tests
Logic verification or functionality tests
- Done before chip tapeout to verify the functionality of the
design
Silicon debug
- Run on the first batch of chips that return from fabrication
- Confirm that the chip operates as intended and help
debug any discrepancies
- Can be done at full speed, more extensive than logic
verification tests
- Less visibility inside the chip compared to design phase
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
4
Main Categories of Tests
Manufacturing test or Production test
- Done on each manufactured chip before shipping to the
customer to verify that the silicon is completely intact
- Verify that every transistor, gate, and storage element in
the chip functions correctly
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
5
Yield
Not all die on a wafer function correctly
There may be bridged connections or missing features due
to dust particles, imperfections in materials or
photomasking, etc.
Imperfections may result in a fault
Yield is the number of good die divided by the total number
of die per wafer
The goal of a manufacturing test procedure is to determine
which die are good and should be shipped to customers
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
6
Cost of Detecting a Fault
Wafer:
$0.01 - $0.1
Packaged chip:
$0.1 - $1
Board:
$1 - $10
System:
$10 - $100
Field:
$100 - $1000
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
7
Exhaustive testing
n
Department of Computer Systems / TKT-1527
Combinational
logic
Design for Testability / O. Vainio
7.7.2015
8
clk
reg
m
m
n
Combinational
logic
Example:
n = 25
m = 50
Test takes 1 ms per pattern
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
9
Basic Digital Debugging
When a chip returns from fabrication, the first tests are run
in a lab environment
Need a circuit board with features to support testing:
- Power for the IC with ability to vary VDD and measure
power dissipation
- Analog and digital inputs and outputs as required
- Clock inputs as required
- A digital interface to a PC
- Zero insertion force socket for the chip
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
10
The chip should have a serial UART port or some other
interface that can be used independently of the normal
operation of the chip
Software should provide for peeking (reading) and poking
(writing) registers in the chip
Also an interface for a logic analyzer may be provided
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
11
Initial Steps in Chip Testing
”Smoke test”: ramp up supply voltages from zero without
clocks running and monitor the current.
Enable the clocks, some dynamic current should be
evident. If possible, initially use reduced clock speed.
Examine various registers using PC-based peek and poke
software, checks the integrity of the interface.
The chip may have built-in self test that can be activated
over a boundary scan interface.
Otherwise, the functionality is checked from the bottom-up.
Top-level test like running a piece of code at once often
does not work, usually because of problems with the test
fixture.
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
14
Testers and Test Fixtures
A tester is a device that can apply a sequence of stimuli to
a chip or system under test and monitor and/or record the
results
Four general types of test fixtures are:
- A probe card to test wafers or unpackaged dies
- A load board to test a packaged part
- A PCB for bench-level testing (with or without tester)
- A PCB with the chip in situ, demonstrating the application
for which the chip is used
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
15
Handlers
An IC handler feeds ICs to a test fixture attached to a
tester.
Devices are gravity-fed to a handler, which mechanically
picks the chips up and places them in the test socket on
the load board.
The tester stimulus is then applied and chips are binned
depending on whether or not they passed the test.
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
16
Shmooing
The ability to vary the voltage and timing on a per-pin
basis with a tester allows a process called ”Shmooing”.
For instance, vary VDD from 3 V to 6 V while varying the
tester cycle time.
A shmoo plot shows the sensitivity of the part with respect
to voltage.
Another example of shmoo is to skew the timing on inputs
with respect to clock to look for setup and hold variations.
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
17
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
*
*
*
*
*
*
1.0
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
1.1
1.2
1.3
1.4
1.5
voltage
Normal, well-behaved shmoo
Typical speedpath
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
* indicates failure
7.7.2015
18
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
*
*
*
*
*
*
*
*
1.0
*
*
*
*
*
1.1
*
1.2
*
*
*
1.3
1.4
*
1.5
voltage
”Brick wall”
Bistable initialization
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
19
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
1.0
1.1
*
*
*
*
*
*
1.2
*
*
*
*
*
*
1.3
*
*
*
*
*
*
1.4
1.5
voltage
”Wall”
Fails at a certain voltage (coupling, charge sharing, races)
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
20
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
1.0
*
*
*
*
*
*
*
*
*
*
1.1
1.2
1.3
1.4
*
*
*
*
*
*
1.5
voltage
”Reverse speedpath”
Leakage
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
21
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
*
*
1.0
*
*
1.1
*
*
1.2
*
*
1.3
*
*
1.4
*
*
1.5
voltage
”Floor”
Works at high but not low frequency (leakage)
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015
22
Shmoo plot
Clock
Period
in ns
1.0
1.1
1.2
1.3
1.4
1.5
*
1.0
*
1.1
*
1.2
*
*
*
1.3
1.4
1.5
voltage
”Finger”
Coupling
Department of Computer Systems / TKT-1527
Design for Testability / O. Vainio
7.7.2015