Test Pattern Generator
Chapter 17: Analog Test Bus Standard
17.1 Inductance measurement
φ = LI, V = LdI dt
One test method would be to apply a current ramp to the inductor while mea-suring the voltage across it, which should be constant. We will ignore the startup and shutdown transients and just look at the steady-state response.
Since all of the switches induce a voltage drop (due to their 100Ω resistance), we can first disable the inductor by applying a known DC current and measuring the voltage drop in the system. Then, we can calculate the efective series resitance:
V = IRef f, so, Ref f = I V For the current ramp test,
V = Ref fI(t) + LdI dt Let I(t) = 0 + 50 mA/s× t, so dIdt = 50 mA/s. Then,
V (t) = Ref f ×50 mA
s t + L×50 mA s So,
L = V (t)− Ref f50mAs t 50 mA/s or
L = V (t)
50 mA/s − Ref ft Test method (see Figure 17.8, page 586 of the book):
I. Measure Ref f
A. Turn on S5, S6, SB1(pin1), SB2(pin1), SG(pin1), and force 100µA of current into AT 1 with ATE current source.
B. Measure voltage V1 between AT 2 and ground.
C. Turn off SB2 at pin1 and turn on SB2 at pin2.
D. Measure voltage V2 between AT 2 and ground.
E. V = V1 − V2. If this is not 0, then there is significant current leakage through AT 2 or there is series resistance in the inductor.
Ref f = V1− V2
100µA We assume the latter.
II. Measure L
A. Turn off SB2 at pin2 and turn on SB2 at pin1.
B. Apply this current ramp through AT 1:
t slope=50mA/s
Current, I
0
C. Measure V at two different time instants t1 and t2 (chosen so that the startup transient at t = 0 has died down.) Use steps I, B through D, to do this.
D. Compute L twice:
L = V (t)− Ref f50mA s t 50mA/s
The measurements L1and L2should agree within 60×10−10H. If they do not agree within that tolerance, reject the inductor. Otherwise, average the measurements:
Lave = L1+ L2 2
Verify that 59.994µH ≤ Lave ≤ 60.006µH. If Lave is not in this range, reject the inductor.
An error Verr(t) in the V (t) measurement changes the calculated L by
∆L =± Verr(t) 50mA/s
Measurement error exceeds inductor tolerance. This is an unreliable test! We need to change the current slope.
To have |∆L| ≤ 1260µH× 10−4, we need dI
dt = 20µV
|∆L| = 20µV
1
260µH× 10−4 = 6666.67 A/s
Redesign the test with this dIdt, and with sampling times t1 and t2 so that we only get 100mA maximum current.
t1, t2< 100mA
6666.67 A/s = 0.150µs
This may or may not cause a problem with the startup transient being sampled.
Alternative solution:
Apply a sinusoidal current waveform to the inductor and measure the voltage across it, which should have some phase shift. Ignore startup and shutdown tran-sients and just look at the steady-state response.
We assume that the system voltmeter has extremely high impedance, so that we can ignore the current through it and, therefore, we do not model AB2 R0s. Then:
V1 = I(100Ω + jωL) V2 = I(100Ω) V1− V2 = jωLI
Z = jωL =√
ω2L2= ωL L = V1− V2
ωI
Test method (see Figure 17.8, page 586 of the book):
I. Repeat prior part I to calculate Ref f (series resistance of inductor.) II. Measure L
A. Turn off SB2 at pin2 and turn on SB2 at pin1.
B. Apply I = sin(2π× 60t) through AT 1.
C. Measure V1 and V2.
D. V1− V2 = I(Ref f + jωL), ω = 2π60 rad/s
Z = Ref f + jωL =qRef f2 + ω2L2 So,
(V1− V2)2
I2 = R2ef f + ω2L2 L = 1
ω
s(V1− V2)2
I2 − Ref f2
Both Ref f and V1−V2 have an error term proportional to voltmeter error.
17.2 Capacitance measurement
Q = CV, I = CdV dt
One test method would be to apply a voltage ramp to the capacitor while mea-suring the current across it, which should be constant. We will ignore the startup and shutdown transients and just look at the steady-state response.
Since all of the switches induce voltage drop (due to their 100Ω impedance), we can first disable the capacitor by applying a known DC voltage and measuring the current in the system. Then, we can calculate the effective series resistance.
I = V
Ref f so Ref f = I V For the voltage ramp test,
V = IRef f + Z t
0
Idt
C or
dV
dt = I
C (since I will be constant) So,
C = I
dV /dt, choose dV
dt = 50mV /s Test method (see Figure 17.8 on page 586 of the book):
I. Optionally measure Ref f (not needed for accurate C measurement.) Method is the same as for Part I of solution to Problem 17.1.
II. Measure C
A. Turn on S5, S6, SB1 (pin 1), SG (pin 1), and SB2 ant pin 1.
B. Turn off SB2 at pin 2.
C. Apply the voltage ramp, shown below, at AT 1.
0 t
slope=50mV/s
Voltage, V
E. Compute C twice:
C = I
dV /dt
The measurements should agree within 0.02nF . If they do not, reject the capacitor. Otherwise, average the measurements.
Cave = C1+ C2
199.98nF ≤ Cave≤ 200.02nF If not, reject the capacitor.
Alternative solution:
Apply a sinusoidal current waveform to the capacitor and measure the voltage in the circuit, which should have some phase shift. Ignore startup and shutdown transients, and just look at the steady state response. Ignore impedances of S6 and SB2 in AT 2, since the voltmeter has very high impedance.
V1 = I(100Ω + 1 jωC) V2 = I(100Ω)
V1− V2 = I jωC
Z = 1
jωC = jωC
−ω2C2 =
sω2C2 ω4C4 = 1
ωC
C = I
ωC(V1− V2)
Test method to measure C (see Figure 17.8 on page 586 of the book):
A. Turn off SB2 at pin 2 and turn on SB2 at pin 1.
B. Turn on S5, SB1 at pin 1, and SG at pin 2.
C. Apply I = sin(2π× 60t) through AT 1.
D. Measure V1 and V2,
Z = 1
ωC, so (V1− V2)2
I2 = ω2C2 C =
s(V1− V2)2 ωI
17.3 Resistance measurement
To measure a resistance R of value 40Ω±10%, we proceed as follows (see Figure 17.8 on page 586 of the book):
A. Turn on S5, S6, SB1 (pin 1), SB2 (pin 2), SG (pin 1) and force 100µA of current into AT 1 with ATE current source (AT 1 connects to AB1, AT 2 connects to AB2.)
B. Measure voltage V1 between AT 2 and Ground.
C. Turn off SB2 at pin 1 and turn on SB2 at pin 2.
D. Measure voltage V2 between AT 2 and Ground.
E. V = V1− V2 and Ref f = V1− V2. This will be correct only if the ATE system voltmeter has extremely high impedance (so that current flow in AT 2 can be ignored.) If that holds, there is no voltage drop in SB2 at pins 1 and 2 and at S6 in AT 2. However, notice that the external R that we are measuring (40Ω) is smaller than 100Ω, the switch resistance.
First, we calculate the effect of voltmeter measurement error of 20µV : V1ef f = V1+ 20µV
V2ef f = V2− 20µV
Ref f = V1± 20µV − (V2∓ 20µV ) 100µA
= V1− V2
100µA ± 0.4Ω
The error of 0.4Ω equals the tolerance on the resistor value of 40Ω. Although this test can be used, it will most likely reject the vast bulk of resistors tested. A better test would be to reduce the ATE system voltmeter measurement error to 2µV . Alternatively, the circuit under test can be redesigned not to use such low-valued R’s.
We should now calculate a bound on the leakage current Ileak in the voltmeter on AT 2.
V1AT 2 = V1− (RSB2p1+ RS6)Ileak V2AT 2 = V2− (RSB2p2+ RS6)Ileak
V1AT 2− V2AT 2 = V1− V2− Ileak(RSB2p1+ RS6− RSB2p2− RS6) V1− V2 = V1AT 2− V2AT 2− Ileak(RSB2p1− RSB2p2)
Error = ±Ileak(RSB2 mismatch between pins 1 and 2)
RSB2 mismatch Upper bound on Ileak
10Ω 0.2µA
5Ω 0.4µA
2Ω 1.0µA
1Ω 2.0µA
0.1Ω 20.0µA
17.4 Analog core test
The difference between INTEST and PROBE instructions for testing an analog core is as follows. PROBE does not require boundary scan hardware on lines from digital circuits to analog circuits, whereas INTEST does. As a result, the PROBE instruction lets internal digital circuits interact with the analog core, while INTEST disconnects the digital core and replaces it with set up patterns from the boundary register.
17.5 Bus impedance calibration
Measurement of resistive impedance in AT 1 and AT 2 buses:
1. In the test bus interface circuit (TBIC), close switch S1 connecting AT 1 to VH. Ground the other end of AT 1 at the automatic test equipment (ATE).
Measure the current in AT 1, I = (VH− 0)/RAT 1. So, RAT 1 = VH/I.
2. Repeat step 1, but using AT 2 and closing switch S2. Thus, RAT 2= VH/I.
3. In the TBIC, close switch S3 to connect AT 1 to VL. Set the other end of AT 1 to VDD at the ATE. Measure the current in AT 1, I = (VDD− VL)/RAT 1. So, RAT 1 = (VDD− VL)/I. Compare this value to the result of step 1.
4. Repeat step 3, but using AT 2 and closing switch S4 in the TBIC. Thus, RAT 2 = (VDD− VL)/I. Compare this value to the result of step 2.
An alternative procedure:
1. Disconnect all pins from AB1.
2. In the TBIC. close switches S5 and S8 to connect both AT 1 and AT 2 together through AB1.
3. Drive AT 1 with a voltage V1 and drive AT 2 with V2. Measure the current I flowing into AT 2 at the ATE:
I = V1− V2
Rtotal So,
Rtotal= V1− V2
I
17.6 Bus impedance calibration
1. First, measure resistance in AT 1 and AT 2 switches S5 through S8.
2. At a given pin, connect AB1 by closing SB1 and close SG for that pin. All other pin switches should be open. Connect AT 1 to AB1. Force VDD into AT 1 at the automatic test equipment (ATE). Measure I at AT 1:
ATE AT1 S5 AB1 SB1
SG VDD
VG TBIC
Analog pin
I = VDD− VG
RS5+ RSB1+ RSG So,
RSB1+ RSG= VDD− VG
I − RS5
Assume:
RS5= RS5+ RS8
2 , from Problem 17.5 3. Repeat step 2 closing SL at the pin rather than SG. Then,
RSB1+ RSL= VDD− VL
I − RS5
4. Repeat steps 1 through 3, but using AB2 instead of AB1 and RS7rather than RS5.
Alternative method:
1. Probe the analog pin with the ATE.
2. Measure resistance in AT 1 and AT 2 switches S5 through S8.
3. At the pin, connect AB1 by closing SB1 for that pin. All other pin switches should be open. Connect AT 1 to AB1. Force VDD into AT 1 at the ATE.
Make the ATE force the analog pin to ground. Measure I at AT 1.
AB1 S5
AT1 SB1
VDD Analog pin
I = VDD RS5+ RSB1 RSB1 = VDD
I − RS5
RS5 = RS5+ RS8
2 , from Problem 17.5
4. Repeat step 3, but using AB2 instead of AB1 and RS7 rather than RS5.