The Practical Measurement Setup of DPI Method

above 1 GHz for ICs

Yin-Cheng Chang1,2, Shawn S. H. Hsu2, Member, IEEE, Kang-Chu Peng1, Yen-Tang Chang3, Chiu-Kuo Chen3,
Hsu-Chen Cheng1, and Da-Chiang Chang1
1
National Chip Implementation Center, National Applied Research Laboratories, Hsinchu, Taiwan
2
Institute of Electronics Engineering, National Tsing Hua University, Hsinchu, Taiwan
3
Bureau of Standards, Metrology and Inspection, M.O.E.A, Taipei, Taiwan

Abstract—A reliable DPI measurement is proposed to investigate

the IC immunity above 1GHz. The direct RF power injection
(DPI) method is reviewed, and the concern for extending
frequency range is discussed. Details of the measurement setup
are reported in this work. The critical part, on-board injection
network, in the power injection path is realized with a 3dB
bandwidth of 14.5GHz. A low dropout regulator is used to
demonstrate the test and setup. The proposed DPI measurement
above 1GHz is validated by the experimental results.

I. INTRODUCTION
Moore’s law increases the significance of the research of
the electromagnetic compatibility (EMC) of integrated
circuits (IC) because of the continuous miniaturization of the
Fig. 1 The test setup of DPI method for ICs.
feature size in IC technology. Scaling down the size of the
devices as well as the increasing transistors amount allows IC Furthermore, the models of DPI test setup were built for
high-speed operation driven by lower power. The simulation which agrees well with measured results [7-8].
consequently desired high performances in the opposite side Based on the mature modeling, the differences between DPI
not only produce noise but also make the IC itself vulnerable and BCI (bulk current injection) [9] tests are analysed and the
to interference. It leads the demand of characterizing their high conformity is observed [10]. These valuable literatures
behaviours of emission and immunity. Therefore, several provide informative issues for DPI and polish it. The rest part
measurement methods have been developed as the standards. of DPI to be looked for is its applicable frequency bandwidth.
The technology subcommittee 47A of International Most released standards like IEC series have the
Electrotechnical Commission (IEC) published a series of frequency range below 1 GHz. And sometimes it is
chip-level test methods for EMI (61967 series) [1] and EMS insufficient to evaluate the EMC behaviours while the modern
(62132 series) [2]. They are widel y adopted as the circuits operate higher than 1 GHz. Therefore, some
comparative evaluation for choosing the best candidate of measurement methods like GHz transverse electromagnetic
product from different designs. Among them, a method to (GTEM) cell [11] was proposed which has the frequency
measure the immunity of ICs called direct RF power injection range up to 18GHz. As the trend, the DPI is also expected to
(DPI) [3] is popular as shown in Fig. 1. The straightforward having the capability of a wider bandwidth. [12] proposed a
test setup helps to investigate the conducted emission new methodology by using the edge coupled transmission line
behaviour of the certain IC pin/pins. Relying on the help of as a part of the injection network. The result shows
measurement, most EMC performance can be classified. [4] requirement above 1GHz can be achieve, but the frequency
utilized the DPI to differentiate the improved susceptibility bandwidth was limited by the narrow band nature of coupler.
levels of an IC with several embedded on -chip EMI This paper is organized as follows. At first, the DPI method
protection. Similarly, the ESD protection strategy uses DPI as is revisited and some principles are emphasised. In section III,
the EMI aggression to demonstrate the impact on EMC the feature of DPI test above 1GHz is proposed. The
performance [5]. Furthermore, DPI was applied to help build measurement setup and discrete components used in the
the model of a high-voltage p-channel metal-oxide- injection path with the test board are discussed in details.
semiconductor (HV-PMOS) which can predict accurately the Finally, a LDO used widely in communication module is
susceptibility in transistor level [6]. These reveal the fact that tested as the DUT. The result of immunity level from the
DPI is recommended when the experimental results were conducted RF disturbances is measured, and the measurement
desired to be reproducible, repeatable, and confident. setup is validated.
 0.0 DC supply

-0.2 RF Choke
RF
Injection
-0.4
S21(dB)

DC+RF
DC block to IC
-0.6
(a) (b)
Fig. 4 (a) The configuration and (b) photograph of on-board injection
-0.8 network

directional coupler can obtain the reflected power. That is

-1.0 because different DUTs have different impedances, matching
0 2 4 6 8 10 12 14 16 18 20
Freq (GHz) situations at their pin under test. To differentiate the immunity
levels of ICs, the Pforward is more meaningful. The Pforward is
Fig. 2 The insertion loss of the chosen capacitor.
measured by a power meter with a power sensor. The
1000 dynamic range and frequency range of power sensor should
900 be taken care. Also the VSWR is desired to be smaller than
800 1.15. Notice that the Pforward measured by power meter has to
700
be corrected by adding the coupling factor of the directional
coupler. Accompanying with applications in high frequency,
600
most power injection parts can be found with expected
Z(ohm)

500 performance. Dislike the discrete components in the on-board

400 section of power injection part which will be discussed in the
300 next section.
200 An oscilloscope, test receiver or other monitoring device
100 is used to monitor the malfunction of the DUT during the
0 experiment. The injected power when DUT becomes
0 2 4 6 8 10 12 14 16 18 20 susceptible has to be recorded. A control unit or program can
Freq (GHz)
be used to control these equipments which will save time.
Fig. 3 The impedance of the chosen inductor .
III. ON-BOARD INJECTION NETWORK
II. CONSIDERATIONS OF THE DPI TEST SETUP ABOVE 1GHZ The RF power is injected to the pin of DUT on the PCB.
The DPI method is defined to measure the immunity of IC The traces on PCB have to be design as short as possible with
in the presence of conducted RF disturbances. This conducted a characteristic impedance of 50Ω. A DC block capacitor is
forward power which delivers to a circuit by the cable harness inserted to prevent DC going to the amplifier that may destroy
or the traces on a PCB can be measured. To characterize the it. A 6.8nF is mentioned in the standard which give the lower
immunity of an IC, the forward power that causes malfunction frequency limit around 150 kHz. The larger capacitance can
is measured. The general test setup accordance with the IEC achieve lower 3dB bandwidth and present a high pass
62132-4 standard is shown in Fig. 1. It contains the DC power response. The problem is the parasitic effect makes the
supply, RF power injection part, test PCB with injection resonance happen and make the capacitor become inductive
network links to DUT, monitoring device, and a control unit. and the upper 3dB roll-off occur. Therefore, a capacitor with
Several elements in the power injection part become wide bandwidth, flat frequency response, and low insertion
critical while the measurement frequency extending above loss is preferred. In this work, a 100nF capacitor (ATC 545L)
1GHz. In order to deliver enough power into DUT pin, the is chosen with its S-parameters can be obtained for estimation
50Ω characteristic impedance (Z0) system has to be in advance. Fig. 2 shows it has a low insertion loss over
implemented for effective power level and less path loss. The 20GHz.
power level from signal generator is often insufficient at high If the pin under test is also supplied by a DC source, a
frequency. Therefore a power amplifier is driven for decoupling component is necessary to avoid the injected RF
providing enough disturbance level into the pin under test power heading to the DC source where presents a low
which often presents high degree of mismatch. Because of the impedance AC path. Generally, a RF choke like inductor is a
requirement of varied frequency, sometimes several good candidate. It is recommended to have the AC impedance
amplifiers are needed to cover a wideband measurement. over 400Ω in the test frequency range and not cause too much
Besides, the level of harmonics has to be 20dB lower than DC voltage drop on the path. Again, the parasitic has to be
interference according to the standard. minimized that can guarantee higher operating frequency
The directional coupler is employed to monitor the power without resonance. A 2uH inductor (ATC 506WLS) is
(Pforward) injected to the port of test PCB from power amplifier. selected, and the impedance as a function of frequency is
Only the P for ward is concerned, though another port of shown in Fig. 3. The AC impendence is greater than 360Ω
 100
90
80
70
60
Z0(ohm)

50
40
30
20
10
0
0 1G 2G 3G 4G 5G 6G
Freq (Hz)
Fig. 5 Verification of Z0 for 1cm trace on PCB.

0 Fig. 7 The photograph of DPI test setup above 1GHz.

S21_s2p combination
-10 S21_injection network
S11_injection network
Magnitude(dB)

S11_s2p combination

-20

-30

-40 (a) (b)

0 2 4 6 8 10 12 14 16 18 20 Fig.8 (a) The configuration and (b) photograph of test board.
Freq (GHz)
Fig. 6 The S-parameters of on-board injection network. can be delivered into DUT effectively. The following section
will apply this injection network to achieve a DPI testing.
within 20GHz bandwidth. Both Fig. 2 and 3 are the ideal
response while the component stands alone, the response will IV. EXPERIMENT OF DPI METHOD ABOVE 1GHZ
get worse when in conjunction with other components. The test setup for DPI measurement follows Fig. 1, it
The capacitor and inductor are formed a bias tee as an contains the DC power supply, power meter, RF power
injection network as shown in Fig. 4(a). They are mounted on injection part (RF generator, RF amplifier, directional coupler,
a 1.6mm FR4 double side PCB as shown in Fig. 4(b). The on-board injection network), test PCB with DUT, and the
vector network analyzer used in this work to measure the S- oscilloscope as monitoring device, but lack of the control unit.
parameters is Agilent PNA N5230A with the measurement Fig. 7 shows the photo of the test setup.
capability of 300 KHz to 20 GHz. The standard four-port A low dropout regulator (LDO) TLV70018 for portable
short-open-load-thru (SOLT) calibration was performed devices is chosen as the IC under test. The function of LDO is
before testing. A 1cm coplanar waveguide transmission line to provide an accurate and stable DC voltage to the system.
as the trace on PCB is designed to verify the Z0 of 50Ω as This device is widely used in the modern communication
shown in Fig. 5, which can reduce the mismatch. The modules which operate at the frequency range from hundred
resonance causes the Z0 leave 50Ω at the quarter wavelength MHz to several GHz. So it is a good candidate to perform DPI
corresponding to the frequency around 4-5GHz is observed. test. Fig. 8 shows the test board with injection network and
This notices that the trace for connection on PCB has to be as the typical configuration of IC for general application. The
short as possible. In the standard, a 3dB insertion loss of the injection point is set at the VDD to emulate once the
on-board injection network is permitted to perform the DPI interference injects into it and cause malfunction, the whole
test. Fig. 6 shows the measured S-parameters of the system will fail too.
implemented injection network. The stand alone S-parameters To observe the DUT fails, the output pin is connected to
of capacitor and inductor are combined and simulated to make an oscilloscope (Agilent DSA91204A 12GHz real time
comparison with the implemented case. The discrepancy oscilloscope) through a 1MΩ probe. The immunity criterion
exists in S21 is mainly contributed by the extra SMA has to be defined to tell if the DUT fail or not when subject to
connectors and the parasitic effect at high frequency. The the interference. A failure is defined when the output voltage
measured 3dB bandwidth reveals the DPI measurement reach ±4% tolerance by referring to the data sheet (2%
frequency range can be expended up to 14.5GHz. And the accuracy at 1.8V). In other words, once the output voltage
return loss larger than -10dB confirms that the injected power lower than 1.728V or higher than 1.872V, it fails. No matter
 V. CONCLUSION
In this paper, the setup of establishing DPI measurement
above 1GHz ICs is proposed. The on-board injection network,
in the power injection path is designed and verified. To
achieve such a wide bandwidth measurement, all the
components, instruments, and Z0 of the PCB traces are
carefully considered. By employing a LDO as the DUT, this
proposed DPI measurement above 1GHz is validated. The
experimental result shows the capability of investigating the
immunity of ICs up to 8GHz.
ACKNOWLEDGMENT
The authors would like to thank research group from
BSMI (Bureau of Standards, Metrology and Inspection) for
Fig.9 The measured waveform when DUT reach the immunity criterion. their technical support. The original research work presented
40 in this paper was made possible in part by the BSMI under
Contract No. 2C101011222-05, grant from BSMI, Taiwan.
35
30 REFERENCES
[1] Integrated Circuits, Measurement of Electromagnetic Emission, 150
25
Pforward(dBm)

KHz to 1 GHz: General Conditions and Definitions—Part 1,

International Electrotechnical Commission Standard IEC61967-1, Mar.
20
2002.
15 [2] Integrated Circuits, Measurement of Electromagnetic Immunity, 150
KHz to 1 GHz: General Conditions and Definitions—Part 1,
10 International Electrotechnical Commission Standard IEC62132-1,
2007.
5 [3] Integrated Circuits, Measurements of Electromagnetic Immunity 150
kHz to 1 GHz—Part 4: Direct RF Power Injection Method, Standard
0
0 1 2 3 4 5 6 7 8 9 IEC 62132-4, 2006.
Freq (GHz) [4] A. Alaeldine, N. Lacrampe, J. L. Levant, R. Perdriau, M. Ramdani,
and F. Caignet, “Efficiency of embedded on-chip EMI protections to
Fig.10 The immunity of LDO by applying DPI measurement above 1GHz continuous harmonic and fast transient pulses with respect to substrate
injection,” in Proc. IEEE Symp. EMC 2007, Jul. 9–13, pp. 1–5.
the average value or the AC ripple touch this limit, the Pforward [5] K. Abouda, P. Besse, and E. Rolland “Impact of ESD strategy on EMC
is recorded and represents the immunity level at that performances,” in Proc. International Workshop on Electromagnetic
frequency. Fig. 9 displays an example waveform while failure Compatibility of Integrated Circuits EMC Compo, Nov. 2011, pp. 224-
229.
occurs at 1GHz. Besides, the upper injected power level has [6] O. Jovic, C. Maier, A. Baric, “High-voltage PMOS transistor model
to be set based on the performance of facilities. In this work, a for prediction of susceptibility to conducted interference” IEEE
maximum Pforward of 30dBm is defined because of the Transactions on Electromagnetic Compatibility., Vol. 53, Feb. 2011,
restriction of instrumentation. The DPI test was demonstrated pp. 53–62.
[7] Loeckx J., Georges G. “Assessment of the DPI standard for immunity
from 1 GHz to 8 GHz with a frequency step of 500 MHz. Fig. simulation of integrated circuits,” IEEE Symp. on EMC: Workshop and
10 illustrates the result of a DPI measurement. The immunity Tutorial Notes, Honolulu, HI, USA, 2007, vol. 1–3, pp. 813–817.
remains at 30 dBm limit after 3GHz indicates the pin under [8] A. Alaeldine, R. Perdriau, M. Ramdani, J. Levant, and M. Drissi, “A
test is susceptible to the electromagnetic aggression because direct power injection model for immunity prediction in integrated
circuits,” IEEE Trans. Electromagn. Compat., vol. 50, no. 1, pp. 52–62,
of its robustness. All the details of used instruments and Feb. 2008.
components are listed in Table. 1. [9] Integrated Circuits, Measurements of Electromagnetic Immunity 150
TABLE I kHz to 1 GHz—Part 3: Bulk current injection (BCI) method, Standard
INSTRUMENTS AND C OMPONENTS USED IN DPI SETUP ABOVE 1GHZ IEC 62132-3, 2007.
Component/Instrument Vendor / part Feature [10] S. Miropolsky, S. Frei “Comparability of RF Immunity Test Methods
RF generator Agilent/E8247C 250k~20GHz for IC Design Purposes,” in Proc. International Workshop on
RF amp Mini-Circuits ZVE-3W 2G~8GHz Electromagnetic Compatibility of Integrated Circuits EMC Compo,
Nov. 2011, pp. 59-64.
Directional Coupler Agilent 87300B 1G~20GHz
[11] D. Konigstein and D. Hansen, “A new family of TEM-cells with
Power sensor Agilent E4413A 50M~26.5GHz enlarged bandwidth and optimized working volume,” Proc.7th Int.
Power meter Agilent E4416A 20M Sa/sec Zurich Symp. and Tech. Exh. on EMC, pp. 127 – 132, March 1987.
Oscilloscope Agilent DSA91204A 12GHz, 40G Sa/sec [12] J. Catrysse, D. Pissoort, and F. Vanhee, “Expanding the frequency
DC supply Agilent E3615A 0~20V, 0-3A range for DPI testing of IC’s above 1 GHz: an alternative proposal,” in
Capacitor ATC 545L 100nF, 16kHz~40GHz Proc. Int. Symp. Electromagn. Compat. (EMC Europe), Sep. 2011, pp.
400–404.
Inductor ATC 506WLS 2uH, 400kHz~40GHz
LDO TLV70018 1.8V output