How To Use Picoprobes And Flexible Pitch Probes

How To Use Picoprobes And Flexible Pitch Probes
Advanced TLP/HMM Solutions
This application note describes the four point Kelvin
standard TLP method with discrete current sensor, fixed
pitch and variable pitch wafer level probe tips.
L is the length of the 50 Ω interconnection cable between the transformer-based current sensor and the
DUT. The propagation velocity of the pulse signal in the
50 Ω cable is approximately
c
v = √ ≈ 0.2 m/ns
r
1 Four Point Kelvin TLP Method
Fig. 1 shows the four point Kelvin TLP method including dc leakage measurement. This setup is widely used
for standard TLP measurements with high accuracy because of separated pulse force and pulse sense channel. A transformer-based pulse current sensor (e.g.
Tektronix CT-1) is used in the pulse force channel. A
tiny 4950 Ω resistor which is integrated in front of the
pulse sense probe tip acts as voltage divider 100 : 1 =
(4950 Ω + 50 Ω)/50 Ω together with the 50 Ω transmission line and 50 Ω input impedance of the digital oscilloscope.
where speed of light c ≈ 2.998 · 108 m/s and r ≈ 2.1 is
relative dielectric constant of PTFE (Teflon) based insulator of the 50 Ω coaxial cable.
The readout of the current sensor is wrong in
the first time up to:
t = 2·
GPIB
SMU
DUTILeakageITest
t
L
...
...
v
...
Leakage
DUT
GPIB
Transformer4Based
PulseICurrentISensor
SwitchIControl
DUTIPulseIForce
Pulse
5- Ω PulseIOutput
5- Ω
DUT
SWITCH
PAD
PAD
DUTIPulseISense
VDUTI(t)
GPIB
IDUTI(t)
time in [ns]
distance between current sensor
and DUT in [m]
0.2 m/ns pulse signal propagation
velocity
1. Propagation velocity in the cable is v ≈ 0.2 m/ns,
according Eqn. 1
4995IkΩ
5- Ω
PulseISense
(2)
Let’s investigate the following example test conditions:
5- Ω
DIGITAL
OSCILLOSCOPE
L
v
5- Ω Picoprobe
ModelIy-
LI<I-9yIm
5- Ω
5- Ω
PC + System Control Software
LEAKAGE
TEST
TLP
GENERATOR
(1)
5IkΩ Picoprobe
ModelIy-
2. Distance ”L” from DUT to current sensor (Fig. 2)
TinyIresistorIbuiltIinIfrontIof
theIpulseIsenseIprobeItip9
Current
5- Ω
Ay
3. Rise time of the TLP is set to 100 ps and pulse
width is set to 100 ns
Figure 1: Four point Kelvin TLP method including dc
leakage measurement.
4. TLP voltage is set to VPulse = 1000 V
5. Device-under-test (DUT) is a 5 Ω resistor
1.1 Why the current sensor should be
connected to the DUT as close as
possible?
For this load condition we expect
Fig. 2 is a simplified view of the block diagram shown in
Fig. 1. The probe tips with equivalent circuit are shown
in Fig. 9.
TLPtGenerator
PulsetSignal
Propagation
Transformer-Based
PulsetCurrenttSensor
L
50 Ω
50 Ω
DUT
PAD
PAD
IDUT
VDUT
DUTtPulsetSense
50 Ω
5tkΩ Picoprobe
Modelt10
4.95tkΩ
Tinytresistortbuilttintfronttof
thetpulsetsensetprobettip.
Figure 2: TLP generator, pulse force channel, pulse
sense channel and current sensor output.
Application Note AN-010
Rev. 1.4
September 4, 2014
IDUT
=
VPulse
· 5 Ω = 90.9 V
50 Ω + 5 Ω
VPulse
= 18.2 A
50 Ω + 5 Ω
(3)
(4)
1.1.1 Test Case 1: L = 0 m
50 Ω
DUTtPulsetForce
CurrenttSensortOutput
=
50 Ω Picoprobe
Modelt10
50tΩ
VPulse
VDUT
If L = 0 m, which means the current sensor is mounted
directly at the DUT (which is not possible in reality), we
would measure the following DUT voltage and current
waveforms, as expected by Eqn. 3 and 4, confirmed by
TLP simulation results1 shown in Fig. 3.
Fig. 4 shows the same plot but with higher time resolution up to t = 20 ns. As a summary, with L = 0 m
the voltage and current waveforms occur immediately at
same time t = 0 ns with correct value.
1 TLP
Page 1 of 4
wave reflection simulation including cable losses.
High Power Pulse Instruments GmbH
Internet: www.hppi.de
E-mail: [email protected]
How To Use Picoprobes And Flexible Pitch Probes
100
100
90
90
80
80
VDUT [V], IDUT [A]
VDUT [V], IDUT [A]
Advanced TLP/HMM Solutions
70
60
VDUT
IDUT
50
40
30
70
60
40
30
20
20
10
10
0
-10 0 10 20 30 40 50 60 70 80 90 100 110 120
Time [ns]
VDUT
IDUT
50
0
-2
Figure 3: Simulated TLP DUT waveforms with L = 0 m
condition.
100
0
2
4
6
8 10 12
Time [ns]
14
16
18
20
Figure 5: Simulated TLP DUT waveforms with L = 0.2 m
condition.
readout of the current waveform at t > 2 ns onwards.
90
VDUT [V], IDUT [A]
80
70
1.1.3 Test Case 3: L = 1 m
60
VDUT
IDUT
50
Now we assume L = 1 m to review the expected waveforms in Fig. 6.
40
30
20
100
10
90
80
-2
0
2
4
6
8 10 12
Time [ns]
14
16
18
20
Figure 4: Simulated TLP DUT waveforms with L = 0 m
condition. Detail view up to t = 20 ns.
VDUT [V], IDUT [A]
0
70
60
VDUT
IDUT
50
40
30
20
1.1.2 Test Case 2: L = 0.2 m
10
Now we assume L = 0.2 m, which means the current
sensor is mounted 0.2 m far away from the DUT in
Fig. 2. In this case we would measure the waveforms
shown in Fig. 5.
At t = 0 the propagating pulse signal (v ≈ 0.2 m/ns)
in the DUT pulse force line (Fig. 2) hits the current sensor. The readout of the current sensor at 0 < t < 2 ns
will result into
VPulse
I(t)
=
= 10 A
(5)
2
· 50 Ω
0<t<2 ns
After 1 ns, at t = 1 ns, the ropagating pulse signal
hits the DUT. Therefore immediately the expected correct voltage reading of VDUT = 90.9 V occurs. However,
the propagating pulse signal (the reflected wave) needs
to go back to the current sensor for another 1 ns. Only
if incident and reflected signals are available for superposition at the current sensor, we can expect the correct
Application Note AN-010
Rev. 1.4
September 4, 2014
0
-2
0
2
4
6
8 10 12
Time [ns]
14
16
18
20
Figure 6: Simulated TLP DUT waveforms with L = 1 m
condition.
The correct current readout occurs at t > 10 ns, as
predicted by Eqn. 2.
1.1.4 Test Case 4: L = 1 m and Rise Time tr = 10 ns
Finally, we assume L = 1 m and a TLP rise time of
tr = 10 ns instead of tr = 100 ps of the previous test
cases. In general the same effects occur as shown in
Sect. 1.1.3 but the waveforms are filtered and delayed
by the 10 ns rise time filter, which give a result as shown
in Fig. 7.
Page 2 of 4
High Power Pulse Instruments GmbH
Internet: www.hppi.de
E-mail: [email protected]
How To Use Picoprobes And Flexible Pitch Probes
Advanced TLP/HMM Solutions
tips (see also Fig. 2). On the left sided probe tip clearly
the molded resistor can be identified. The mold compound is a little bit larger in comparison to the right sided
50 Ω probe tip without built-in resistor.
100
90
VDUT [V], IDUT [A]
80
70
60
VDUT (L=1m)
VDUT (L=0m)
IDUT (L=1m)
IDUT (L=0m)
50
40
30
Pulse Sense
20
10
0
-10
Tiny resistor 4950
50
0
10
20
30
40 50 60
Time [ns]
70
80
(molded)
Pulse Force
50
Signal
90 100
GND
Figure 7: Simulated TLP DUT waveforms with
L = 0 m, 1 m and TLP rise time of tr = 10 ns.
1.1.5 Summary Test Cases
The previous investigations show that it is recommended to mount the current sensor as close as possible nearby the DUT. L = 0.1 . . . 0.15 m is a good choice
and compromise. If this is not possible a wrong readout
of the current sensor is expected in the first time defined
by Eqn. 2 and the delay of voltage and current waveform
readout needs to be considered.
Figure 9: Fixed pitch picoprobes. Probing a reference
device on the TLP calibration substrate. Detail
view of Fig. 8 including equivalent circuit of the
probe tips.
3 Flexible Pitch Probe Tips
2 Fixed Pitch Probe Tips
Fig. 8 shows a typical fixed pitch (200 µm)
probetips
from
GGB.
Please
refer
to
http://ggb.com/10.html for the model 10 probe tip
holder and description of replacement probe tips.
http://ggb.com/PdfIndex_files/mod10.pdf
Picoprobe Model 10
(Probetip Holder)
Sometimes the fixed pitch is a limitation in case of devices to be measured changes often and have different
pad pitch and dimensions. For these cases the flexible
pitch setup has been developed.
10-5k(0502)-125-W-1
GF-A
Picoprobe Model 10
(Probetip Holder)
10- 50/30-125-W-1
Pulse Pulse GF-A
Sense Force
GND
Tiny resistor 4950
(molded)
Pulse Sense Probe Tip
10-5K(0502)-125-2-L-200
Figure 10: Flexible pitch setup using GGB 1050/30-125-W-1 (pulse force) and GGB
10-5k(0502)-125-W-1 (pulse sense) replacement probe tips and GF-A flexible pitch
ground fixture.
Pulse Force Probe Tip
10-50/30-125-W-2-R-200
Figure 8: Fixed pitch picoprobes. Probing a reference
device on the TLP calibration substrate.
Fig. 9 shows the detail view of Fig. 8 including the
equivalent circuit of the GS-type (ground-signal) probe
Application Note AN-010
Rev. 1.4
September 4, 2014
GND
The 10-50/30-125-W-1 (pulse force) and 105k(0502)-125-W-1 (pulse sense) probe tips have a
center conductor/needle but no ground needle. The
ground contact is made using a clamp, short flexible
Page 3 of 4
High Power Pulse Instruments GmbH
Internet: www.hppi.de
E-mail: [email protected]
How To Use Picoprobes And Flexible Pitch Probes
Advanced TLP/HMM Solutions
wire, nozzle for a 0.5 mm needle and a separate
probe arm for the ground needle. Despite the ground
connection using a clamp with wire the pulse signal
quality remains very good for rise times down to 200 ps
and below.
Figure 11: Ground fixture GF-A assembly:
clamp
mounted on picoprobes, flexible wire and
nozzle including 0.5 mm tungsten needle.
Figure 13: Please make sure that the probe is clamped
in the groove.
Fig. 11 shows the setup with removed micropositioners for the ground needles.
Figure 12: Ground fixture assembly, two times, for
pulse sense and pulse force: clamp, flexible
wire and nozzle including 0.5 mm tungsten
needle.
Figure 14: Detail view: ground clamp.
Fig. 12 shows the ground fixture comprising the
clamp, flexible wire and nozzle including 0.5 mm tungsten needle. The clamp needs to be fixed on the picoprobe tip. The tungsten needle shall be mounted on a
separate DC probe arm including micropositioner (not
supplied by HPPI). Fig. 13 shows the clamp which has
a groove inside the jaws. Please make sure that the
probe tip is clamped in the groove in order to have stable position and good electrical contact. Fig. 15 shows
the clamp fixed on the probe tip. The final result of the
assembly is shown in Fig. 10
Figure 15: Two ground fixtures GF-A clamped on the
probe tips.
Application Note AN-010
Rev. 1.4
September 4, 2014
Page 4 of 4
High Power Pulse Instruments GmbH
Internet: www.hppi.de
E-mail: [email protected]