THE CHINESE UNIVERSITY OF HONG KONG RFID: Reader Design Cheung Hing Hung 香港中文大學電子工程學系 DEPARTMENT OF ELECTRONIC ENGINEERING
THE CHINESE UNIVERSITY OF HONG KONG
RFID: Reader Design
Cheung Hing Hung
香港中文大學電子工程學系
DEPARTMENT OF ELECTRONIC ENGINEERING
RFID: Reader Design
Author: Cheung Hing Hung
Student I.D.: 02668093
Supervisor: Professor K.K.Cheng
Associate Examiner: Professor K.T.Chan
A project report presented to the Chinese University of Hong Kong
in partial fulfillment of the Degree of Bachelor of Engineering
Department of Electronic Engineering The Chinese University of Hong Kong
April 2006
2
A. Abstract RFID (Radio Frequency Identification), also called electronic labeling, is a new wireless
technology. Reader (Interrogator) can read from or write to a specified identity tag without
any mechanic or optical contact whereas it is done by radio frequency communication.
An RFID system can be simplified into three parts; they are data acquisition, reader and
tag.
In this project, I am responsible for the RFID interrogator and design the system based
on the ISO 18000-6 specification. The interrogator is dedicated to communicate with passive
tags.
Interference is a fatal problem in RFID system, therefore, I designed a technique to due
with it in order to improve the detecting range, and hence, system performance.
Circuit of individual functional blocks, including VTO, AM modulator, power amplifier,
antenna, LNA and mixer, with simulated result and experimental result are evaluated. The
data acquisition capability of a simplified transceiver is also tested.
3
B. Acknowledgements Professor Cheng Kwok Keung, Michael (B.Sc., Ph.D. (London), MIEEE, AMIEE), who
is the supervisor of this project, sets up the goal of this project and gives advice when I faced
some technical problems.
Professor Chan Kam Tai (B.Sc. (Hong Kong), Ph.D. (Cornell), MIEEE) is the associate
examiner of this project.
Kong Cheuk Pang (M.Phil), who is the tutor of this project, helps me to understand the
microwave circuit theory, laboratory instrument, simulation tools and patch antenna theory.
4
C. Content A. Abstract ………………………………………………………………………... P.3
B. Acknowledgements ……………………………………………………………. P.4
C. Contents ……………………………………………………………………..… P.5
D. Introduction………………………………………………………….…………. P.7
E. Theory and Paper Design
1. System specification ……………………………………………………… P.9
2. Estimation of System performance ……………..……………………….... P.13
3. Challenge and solution ……………………………………………………. P.15
4. Materials selection …..…………………………………………………..... P.22
F. Experimental Results
1. Varactor-Tuned Oscillator ……………..…………..……….……..…......... P.24
2. AM modulator …………………………………….…..…………………... P.30
3. Power Amplifier ........................................................................................... P.32
4. Antenna ……………………………………………….…..………………. P.36
5. Variable attenuator …………………………...……...................………….. P.40
6. Voltage Limiter ……………………………………….….………….…….. P.43
7. LNA ……………………………………………………………………….. P.45
8. Mixer ………………………….….....………....…………………………... P.49
5
9. A simplified Interrogator ……….....………....…………………………... P.53
G. Cost Summary ………………………………………….….….……..…………. P.55
H. Discussions and Conclusions …………………………..….…..……………….. P.58
I. Reference ……………………………………………….……...……………….. P.60
J. Appendices
1. Equation of maximum detection range…………………………….…...…. P.61
2. Simulation result of 16-element patch antenna array ……..………….…… P.62
3. Stabilization of HBFP0405 ………………………………………..…. …... P.65
6
D. Introduction The reader is a radio frequency transceiver therefore the design of parameters must fulfill
the radio regulation of local government, on the other hand, International Organization of
Standard (ISO) has designed a clear standard on RFID system, and the name of the document
is ISO 18000-6 [7].
Firstly, one of the specifications is chosen from the document based on the functionality
of desired system. Then, the feasibility and performance of the overall system should be
verified carefully. After it, the specification of individual functional block set can be
confirmed.
Secondly, materials used by the project are selected. The most critical one is apparently
substrate, and other components include microwave transistors, diodes, resistors, inductor,
capacitors, IC chips, SMT connectors and cable.
Firdly, the circuit design of individual block set is concerned. The circuits are tested
under simulation tools like Agilent Advanced Design System (ADS) and Zeland IE3D. ADS
is usually used in transmission line and components related circuit whereas IE3D is used in
Microstrip Antenna design.
7
Fourthly, individually functional block is fabricated and their performance are vertified.
Finally, integrate all the functional blocks into a complete transceiver. The signal
transmission, reception and data acquisition capability are tested.
In this project, I have simulated all the functional blocks that are required in my design
of RFID transceiver. The functional blocks include VTO, modulator, power amplifier, antenna,
LNA, variable attenuator, voltage limiter and mixer. In which, VTO, modulator, power
amplifier and mixer are fabricated. A simplified complete transceiver is also built to test the
data acquisition capability of my design.
8
E. Theory and Paper Design 1. System specification
To begin with, specification of passive backscatter RFID is checked from the ISO
18000-6 document [7]. Table 1.1.1 shows two types of RFID system.
Table 1.1.1
Type
Type B can support more tags and have larger potential in logistic industry therefore it is
selected. There are many parameters but only Modulation index and Data rate is related to my
9
design. They affect the design of modulator and bandwidth of my reader respectively.
Modulation Index
18% modulation index is chosen because it gives a relatively stable power supply to the
passive tag.
Data Rate
10 kbps data rate is chosen because another student, who is responsible for data
acquisition part, plan to build a slow speed system first. In fact, no matter the selected data
rate is 10 or 40 kbps, the turn out bandwidth is still negligible in the radio frequency
transceiver section.
Carrier Frequency
Another important parameter is the carrier frequency. Base on ISO specification, carrier
frequency can be 135 kHz, 13.56 MHz, 433 MHz, 860-960 MHz and 2.45GHz. 860-960MHz
is chosen because antenna gain is higher, size of circuit is more compact and antenna is
smaller at high frequency, but 2.45GHz is not considered because tolerance and side effect
during fabrication can be high for such a high frequency. On the other hand, local radio
regulation further restricts the range to 865-868 MHz & 920-925 MHz, 866 MHz is selected
10
as the carrier frequency in my design finally.
EIRP
By local radio regulation, maximum EIRP is limited by 4W, which is 36dBm. My design
will fully meet this limit in order to provide sufficient power to drive the passive tags.
DC Power Supply
9V battery is used so that the reader can be portable.
The interrogator
Figure 1.1.1 shows a simplified version of the schematic of the interrogator.
Simplified version of the schematic of the interrogator
Figure 1.1.1
11
The transmitting path of the transceiver work as follow:
1. The oscillator gives a power of 10dBm and then it is pre-amplified to 18dBm.
2. Negligible power, about 0dBm, is coupled to the mixer by the non-ideal isolation of a
high impedance capacitor.
3. Most of the power is passed to the AM modulator.
4. The power is further amplified to 26dBm by power amplifier.
5. A quadrature coupler is used to imitate an circulator. The isolated port is connected to
the receiver whereas the two output ports are used to drive an antenna
6. The antenna, with 10dB gain, gives an EIRP of 36dBm, which is the maximum
power limit of a RFID transmitter.
The receiving path of the transceiver work as follow:
1. Received power from the antenna is passed to the quadrature coupler. Half of the
power is lost to the transmiting path and half of the power is passed to the receiving
path
2. Received signal is amplified by LNA, interference are suppressed and output power
is automatically adjusted by AGC before passing to the mixer.
3. Received signal is passed to the mixer and then down-convert to base-band signal
directly.
12
2. Estimation of System performance
Let R be reader, T be tag, d be the distance between the reader and the tag, S be the
received signal power, n be the minimum SNR for detection divided by the isolation between
the transmitting path and the receiving path. The maximum detection range is
πλ
×××
×⎟⎠⎞
⎜⎝⎛=
41)(
41TR GG
ndMax #1 (Proof in Appendix 1)
Table 1.2.1 lists the maximum detection range of the interrogator, based on #1 with
different inputs.
Maximum detection range of the interrogator
GR GT n Max (d)
10 1 1.00E-03 0.49022 m
10 1 1.00E-04 0.87175 m
10 1 1.00E-05 1.55022 m
10 1 1.00E-06 2.75673 m
10 1 1.00E-07 4.90223 m
1 0.1 1.00E-03 0.04902 m
Table 1.2.1
The n parameter is critical to the system performance, a value of 10-3 can be achieved
easily, but a value of 10-6 or 10-7 is challenging. Details will be discussed in next section.
13
The last row of the table shows a bad case with reader antenna gain of 0 dB, tag antenna gain
of -10 dB and n = 10-3, the maximum detection range is about 5 cm only.
14
3. Challenge and solution
RFID system works like a radar system with amplitude modulation, it greatly simplify
the modulation and demodulation processes, especially in the design of tags, it is very
important if a cheap and small tag is desired.
However, interference becomes a big problem because transmitted signal and returned
signal compose of the same carrier frequency; on the other hand, phase is not concerned;
therefore, they are only different in amplitude. Signal must has a larger power than
interference in order to be received.
i.e. where n << 1. Smaller the n, longer the detecting ranges. RPnS ×≥
Figure 1.3.1 shows three sources of interference:
Source of interference
Figure 1.3.1
15
1. Non-ideal isolation of quadrature coupler
2. Reflection from antenna
3. Reflection from external environment
The three interferences listed above have the same frequency but different amplitude and
phase, therefore it is impossible to distinguish between them and eliminate them one by one.
Fortunately, the envelope of mixed interference should be quite stable because the
hardware and external environment will not change rapidly.
Two techniques are required in order to improve detecting range.
1. Interference suppression – if the tag is far from the reader, interference will dominate.
2. Automatic gain control – if the tag is very close to the reader, returned signal will be too
strong and will impact the LNA
Figure 1.3.2 is the schematic circuit I designed to suppress those interferences
16
Schematic diagram of improved system
Figure 1.3.2
Because the envelope of interference varies slowly (say, 0 - 100 Hz) whereas envelope of
wanted signal varies much faster (10 kHz). On one hand, my design tries to suppress all the
signals, which envelope varies slowly, at the input of LNA. On the other hand, wanted signal,
which envelope varies rapidly, can escape from the suppression and can reach the mixer.
The interference suppression loop work as follow:
1. a Wilkinson power divider is used as an adder, in fact, a subtraction
2. Feedback loop with ππ +n2 phase shift is used to allow the feedback signal to
cancell the interference
3. A voltage limiter is used to give a stable power source regardless of the power of the
input signal
4. An envelope detector acquires the output power from LNA, higher voltage implies
17
higher interference
5. The magnitude of canceling signal is controlled by a logic.
Figure 1.3.3 demonstrates how the logic block works. In order to simply the case, I
assume the ππ +n2 phase loop works very well, therefore, the feedback signal can suppress
the interference very well, and then the envelop of interference can be represented by DC
(more accurately, very low frequency signal).
Logic that control the feedback loop
Figure 1.3.3
Starting by the interference at the right top corner, the canceling signal tries to eliminate
the interference, the output will be either
1. canceling signal is too weak, interference still dominates, or
18
2. good canceling, the interference greatly reduced, or
3. canceling signal too strong, it dominates
Passed through the LNA, the signal is strong enough for computing, the envelope of the
signal is regarded as “interference voltage”, after passing through the differential OpAmp, the
output shows the “interference voltage” is rising or falling. If
1. it is rising (say, logic 1), the output of the XOR gate complement
2. it is falling (say, logic 0), the XOR remains the same output
The output of the XOR is charging or discharging a low pass filter (say,τ =0.1) which
gives the canceling signal, therefore, the canceling signal response to the slow changing
interference envelope but not the fast changing data envelope.
Figure 1.3.4 shows a simulation result that use canceling-signal to due with the
interference.
19
Simulating result of interference cancellation
Figure 1.3.4
V1 is input signal (10 kHz Manchester code) plus envelop of interference (imitates by
100Hz sin wave). The interference dominates the signal. If the signal passes through an
Automatic Gain Control, the interference together with the wanted signal will be compressed
this is not desired. Fortunately, with the canceling voltage (V2) keep tracking with the
envelope of the interference, the output remains the wanted signal (Vo) only.
On the other hand, the envelope detector controls the Automatic Gain Control which is
also a voltage controlled attenuator. Unlike the interference canceling system, the AGC
responses to the envelope of the signal instantly when the input signal level is too high.
20
Details of functional blocks of the system will be discussed in experimental results
section.
P.S. White noise is not considered in system design because interference from the transmitter
always dominates.
For example,
Bandwidth = 0.1×Carrier Frequency = MHz 6106.86 ×
T = 300 K
Then,
N = kBT = = 357 fW #2 300)106.86()10374.1( 623 ×××× −
Which is extremely small compared with the interference.
21
4. Materials selection
FR4 is the substrate selected for this project simply because it is cheap. FR4 has rε of
4.3 and cδtan of 0.02. In fact, rε is too low, it makes the size of the circuit and
transmission line quite large; cδtan is high, it is a disaster especially on the efficiency of
patch antenna.
FR4 comes with two substrates’ thickness, 0.8 mm and 1.6 mm. 0.8 mm is used in normal
circuit because it makes the width of transmission line reasonable with respect to the size of
components.
Table 1.4.1 shows the width of strip-line with different substrate thickness whereas the
characteristic impedance is fixed at Ω50 .
Width of strip-line vs substrate thinkness
Thickness Width of strip-line with Ω= 50OZ
0.8 mm 1.54 mm
1.6 mm 3.10 mm
Table 1.4.1
Nevertheless, 3.2 mm thick substrate is used for patch antenna by overlapping two 1.6
mm substrates, this configuration increase the fringe field at the edge, thus, increase
bandwidth and antenna efficiency.
22
For RF transistors and diodes, they should operate up to several gigahertz, several
multiple of the carrier frequency.
In the design of power amplifier, the RF transistors should have high power dissipation.
In the design of low noise amplifier, the RF transistors should have high gain.
In the design of oscillator, the RF transistors should be unstable at the desired frequency.
In the design of variable attenuator, the PIN diodes should have low impedance when
forward current is high; high impedance when forward current is low and small junction
capacitance all the time.
For SMT components i.e. chip resistors, chip inductors and chip capacitors. They should
be sophisticated so that parasitic effect does not happen at the desired frequency. 0603 chip
components, available in the microwave laboratory, are a good choice.
23
F. Experimental Results 1. Varactor-Tuned Oscillator
An accurate oscillator should contain a crystal oscillator, phase comparator,
voltage-controlled oscillator and down converter. In this project, only a varactor-tuned
oscillator is built; whatever, it is sufficient to gives the carrier signal.
There are two main types of oscillator, feedback oscillator and negative-resistance
oscillator. Negative-resistance oscillator is usually used in radio and microwave frequencies.
The design of varactor-tuned oscillator starts from one-port negative-resistance
oscillator.
A negative-resistance device is represented by
),(),(),( ωωω AjXARAZ INININ += where 0),( <ωARIN #3
The load device is represented by
)()()( ωωω LLL jXRZ += #4
The circuit oscillates when
1)(),( =ΓΓ OLOOIN A ωω #5
Which gives
0)(),( =+ OLOOIN RAR ωω and 0)(),( =+ OLOOIN XAX ωω #6
24
Kurokawa [8] has proofed the condition for stable oscillation. It is
0)()()()(>
∂∂
∂∂
−∂
∂∂
∂
==== OOOO
L
AA
INL
AA
IN RA
AXXA
AR
ωωωω ωω
ωω #7
In many case,
0)(=
ωω
ddR L #8
Then, #7 is simplified to
)1()(M
OIN AARAR −−= #9
Power delivered to is LR
)(21]Re[
21 2* ARIVIP IN== #10
By solving dAdP , a convenient value of , which maximizes the oscillator power, is LR
3O
LR
R = #11
The theory of a two-port negative-resistance oscillator is very close to that of a one-port
negative-resistance oscillator. The procedure is:
1. Firstly, use a potentially unstable transistor at the frequency of oscillation.
2. Design the terminating network to make 1>ΓIN . Series or shunt feedback can be used to
increase INΓ i.e. make the RF transistor more unstable at the desired frequency
3. Design the load network to resonate , and to satisfy the start of oscillation condition in INZ
(5.2.22). That is )()( OINOL XX ωω −= and 3O
LR
R = #12
25
n etwork is a com te of strip
igure 2.1.1 is the schematic diagram of the VTO
the VTO is
The RF transistor is at the middle. A tub, at the base of the transistor,
ita t
In order to make the oscillator tunable, the desig of load n posi
line and a capacitor. The capacitor is replaced by a varactor diode finally so that the output
frequency is tunable by varying the property of load network
F
The schematic of
Figure 2.1.1
short circuit s
im tes an inductor to make the transistor more unstable at the desired frequency. On lef
hand side, it is the terminating network. On the right hand side, the load network is a
composite of strip line and capacitor. The bottom-right capacitor will be replaced by a
varactor diode so that the oscillating frequency can be tuned by voltage.
26
Figure 2.1.3 shows the relationship between output frequency and tuning voltage. When
the tuning voltage increased from 0V to 9V, the output frequency increased from 610MHz to
925MHz. At the desired frequency, 866MHz, the tuning voltage is six volts.
Frequency vs Control-Voltage
Figure 2.1.3
28
Figure 2.1.4 shows the output power of the VTO across the tunable spectrum. The output
power of fundamental frequency and first harmonic are 9dBm and -13dBm respectively. The
power difference between fundamental frequency and first harmonic is 21dB, which is good.
The power of fundamental frequency and first harmonic
Figure 2.1.4
29
2. AM Modulator
The AM modulator modulates the carrier with two discrete amplitudes, in fact, it is
simply a variable attenuator switched by PIN diodes. Because 18 % modulation index is
selected, the two amplitude levels have the ratio of 1: 0.835. Figure 2.2.1 shows the layout of
the modulator
Layout of AM modulator
Figure 2.2.1
Control voltages are used to turn on or turn of two pairs of PIN diodes, hence, allow RF
signal to pass through either the lower path of the upper path.
The lower path does not attenuate the signal whereas the upper path gives attenuate the
signal with S21=0.835 by a T attenuator.
30
Table 2.2.1 shows the measurement of the modulator. The reflection coefficient remains
lower than -16dB therefore the circuit is matched to both input and output all the time. The
transmission coefficient has a different of -1.48 dB i.e. 0.843 in magnitude, which is quite
close to desired value, 1:0.835.
Measurement of AM modulator
Digital Siganl High (dB) Low (dB)
S11 -16.69 -20.4
S21 -0.96 -2.44
S12 -0.99 -2.48
S22 -16.57 -20.55
Table 2.2.1
31
3. Power Amplifier
For the design of power amplifier, the choice the transistor is very critical. Unfortunately,
there are limited choices from the market.
The RF transistor being used is BFQ-19S. It has a maximum power dissipation of 1W
whereas my target output power is 400mW.
The bias condition is different from other small signal application. Large
collector-emitter voltage and collector current is selected in order to deliver high power. In my
design, Vce of 8V and Ic of 70mA is used.
In the traditional design of power amplifier, the source and load matching network is
selected to have low impedance values so that the power amplifier gives high output power
provided that voltage or current is fixed. However, since I am going to build a complete
transceiver, an un-matched functional block will affect the properties of other functional
blocks and complicate the whole circuit design, therefore, I chose to match both input and
output port to the transmission line. Fortunately, the performance of the power amplifier does
not deteriorate much.
32
Figure 2.3.2 shows the S-parameters of the power amplifier. The reflection coefficient of
input and output ports are -17.6dB and -16.44dB respectively, therefore they are matched.
S-parameters of power amplifier
Figure 2.3.2
34
Figure 2.3.3 shows the relationship between input power and output power. The power
amplifier has a maximum output power of 24.3dBm and a gain of 9dB.
Output power vs Input power
Figure 2.3.3
The power dissipation of the power amplifier = Vce× I = 7.5×0.1 = 750mW
The maximum RF output power of fundamental frequency = 24.3dBm = 270mW
The efficiency of the power amplifier = 270 / 750 = 36%
35
4. Antenna
As discussed in Materials Selection section, substrate with 3.2 mm thickness and an
antenna of 10 dB gain is required by the system specification. Performance of signal patch
antenna is not satisfactory therefore patch array is used.
Figure 2.4.1 shows the E-field, voltage, current and impedance distribution inside a
rectangular patch antenna.
Properties of single patch antenna
Figure 2.4.1
Length isλ /2 for the desired frequency to resonate. W is optimized to #13 in order to
maximize the gain [4].
2)1(2 +
=r
of
cWε
#13
Most current flow in the middle but not along the edge, therefore an impedance wall is
formed at the edge; proper impedance matching is required in order to feed the patch.
36
Figure 2.4.2 shows a linear 4-element array,
Linear array of four elements
Figure 2.4.2
Length of both patch and feed line areλ /2 so that all patches resonate in phase. From
right to left, the impedance of the feed line is transformed from a high value to a low value,
hence, lower radiating resistance and higher radiating power.
37
Figure 2.4.3 shows a 16-element antenna array; all feed lines are proper designed so that
all rectangular patches resonate in phase.
Patch array of 16 elements
Figure 2.4.3
Simulating result of 16 elements patch array antenna is shown in appendix 2
Table 2.4.1 shows the simulated result of various number of arrays. The efficiency is
limited by the quality of the substrate and the relatively low carrier frequency. The size of the
antenna array can be greatly reduced by using a higher carrier frequency, say 2.4Ghz, whereas
38
the theory of design is almost the same.
The gain of the 16 patches antenna is 1.5 dB above the 10 dB target. The excess gain can
be reserved to compensate other loss in the circuit
Simulating result with different number of elements
number of patches Gain efficiency size
1 ~ 1.5 dB ~ 33 % ~ 11 * 9 cm
4 ~ 6.5 dB ~ 32 % ~ 11 * 65 cm
16 ~ 11.5 dB ~ 31 % ~ 60 * 65 cm
Table 2.4.1
39
5. Variable Attenuator
Bridged-T attenuator [6] is used as the basis of attenuator because only two variable
resistors are required and it is achieved by replacing variable resistors by PIN diodes.
Figure 2.5.1 shows the circuit of a bridged-T attenuator and a voltage-controlled variable
attenuator
Bridged-T attenuator
Figure 2.5.1
Where )110(1 20 −=L
OZR and 110
420 −
= LOZ
R #14
40
In my design, three bridged-T attenuators are cascaded in order to enhance the
performance. The schematic diagram is shown in Figure 2.5.2
Schematic diagram of variable attenuator
Figure 2.5.2
Figure 2.5.3 shows the relationship between transmission coefficient and control-voltage.
Attenuation changes with control voltage. From 0V to 8V, the attenuation varies from
about 20 dB to 80 dB, therefore a dynamic range of 60 dB. The dynamic range implies the
variation of incoming signal that the system can handle. Signals, that out of the dynamic
range, will either saturate the amplifier or not strong enough to be detected.
The minimum attenuator is about 20 dB, it sounds awful. In fact, it will not downgrade
the signal to noise ratio because white noise is negligible. However, a more powerful LNA is
required to boost the power again.
41
S21 vs control-voltage
Figure 2.5.3
Figure 2.5.4 shows the transient response with different control-voltage. The phase
remains constant with different attenuations, it is very important to ensure the ππ +n2
interference suppression loop works.
Transient analysis vs control-voltage
Figure 2.5.4
42
6. Voltage Limiter
This device works like a DC voltage regulator; it gives a constant AC output regardless
of the power of input signal. Figure 2.6.1 shows the schematic diagram of a voltage limiter.
Schematic diagram of the voltage limiter
Figure 2.6.1
43
Figure 2.6.2 shows the simulated result of the voltage limiter. The upper graph is the
input voltage with different magnitude whereas the lower graph is the output. Power is
grounded by PIN diode if the input voltage is either too positive or too negative; therefore, the
output remains almost constant regardless of input power.
Simulating result of voltage limiter
Figure 2.6.2
Besides, the output voltage amplitude depends on the forward threshold voltage of the
PIN diodes.
Again, the output phase remains the same, it is very important to ensure the ππ +n2
loop works.
44
7. Low Noise Amplifier (LNA)
As discussed in Challenge and Solution, white noise is negligible in the system design,
therefore LNA is simply a high gain amplifier.
In this state, the actual gain we required is not confirmed because it depends on the
isolation of interference from LNA and the minimum attenuation of variable attenuator,
however, a high gain amplifier is usually helpful.
In order to obtain high gain and high output voltage swing, two transistors are used. The
first one is HBFP0405, which has very high gain but low output voltage; the second one is
BFR183, which has relatively low gain but high output power. Table 2.7.1 shows the bias
condition of the two RF transistor.
Bias condition of HBFP0405 and BFR183
HBFP0405 BFR183
Vce 2 V 4.5 V
Ic 5 mA 15 mA
Table 2.7.1
Unfortunately, HBFP0405 is unstable from 0 – 7.3 GHz. In order to prevent it from
oscillating, a shunt resistor is added to the output to stabilize the transistor. The
resistor seldom put at the input of the transistor because it will make the output very noisy.
Ω250
45
The corresponding stability diagram is shown in Appendix 3 and the schematic diagram
of a stabilized HBFP0405 transistor is shown in Figure 2.7.1
Schematic diagram of a stabilized HBFP0405
Figure 2.7.1
Table 2.7.2 shows the S-parameters of the two RF transistors after proper biasing and
stabilization.
S parameter of the two transistors
polar HBFP0405 BFR183
S11 0.779 / -40.199 0.091 / -178.41
S12 0.020 / 69.763 0.118 / 72.115
S21 9.569 / 155.77 4.507 / 80.900
S22 0.622 / -13.571 0.391 / -24.389
Table 2.7.2
For HBFP0405, S12 is very low, therefore, unilateral conjugate matching is used, i.e.
571.13622.0199.40779.0
*22
*11
∠==Γ∠==Γ
SS
L
S #15
46
dBS
SS
GTU 8.2586.3791
11
12
22
2212
11max, ==
−−= #16
For BFR183, S12 is quite large; therefore, bilateral conjugate matching is used,
2
22
222
1
21
211
24
24
CCBB
CCBB
ML
MS
−±=Γ
−±=Γ
#17
Where
*11222
*22111
2211
2222
2222
2111
11
SSCSSC
SSBSSB
∆−=∆−=
∆−−+=
∆−−+=
#18
And #19 21122211 SSSS −=∆
2112
2222
211
21
SSSS
K∆+−−
= #20
Solving #6, #7, #8 and #9, finally gives 02.1
9.24783.0181691.0
=∠=Γ∠=Γ
KML
MS
#21
And dBKKSS
GT 1528.31)1( 2
12
21max, ==−−= #22
Cascading the two amplifiers, the total gain = 25.8 + 15 = 40.8 dB #23
Making use of the four values obtained, all ports of the two transistors are matched by
microstrip. Figure 2.7.2 shows the schematic diagram of two-state LNA
Γ
47
Schematic diagram of two-state LNA
Figure 2.7.2
Figure 2.7.3 shows the gain of the two-state LNA verse frequency. The gain is 40.262dB at
the desired frequency, which is very close to the calculated result in #12 that is 40.8 dB.
Gain of LNA verse frequency
Figure 2.7.3
48
8. Single-Balanced Mixer
A mixer is a three-port network with two input, local oscillator and radio frequency, and
one output, intermediate frequency. Mixers are commonly used to multiply signals of
different frequencies in an effort to achieve frequency translation.
A mixer can be divided into three sections. A simple block diagram of a general mixer is
depicted in Figure 2.8.1
Structure of a general mixer
Figure 2.8.1
The first section combine or multiply LO signal with RF signal, in addition, isolation
between LO and RF is very important, otherwise, LO signal will contribute noise to RF signal
or RF signal will distort other functional blocks in the system.
The second section is a non-linear device, which produce multiple frequency
components from the combined signal, LO and RF. Diodes and transistors, with proper
biasing, are commonly used as the non-liner device.
The last section is an IF filter, on one hand, it extracts the desired frequency component
from those frequency components generated by the non-linear device; on the other hand, it
49
suppress unwanted frequency components.
In my design, I chose to use passive single-balanced mixer because, on one hand, it has
higher efficiency than single-ended mixer; on the other hand, the double-balanced and active
properties is not required because the effect of spurious mode is small and the RF signal is
sufficiently strong. Figure 2.8.1 depicts the schematic diagram of the mixer.
Schematic diagram of the mixer
Figure 2.8.1
A quadrature coupler acts as the combiner of LO and RF signal. Because RF and LO signal
compose of the same frequency, a hybrid coupler is an ideal choice to combine the two signals,
in addition, provide excellent isolation between the two input port, LO and RF.
A pair of diodes acts as the non-linear device. The products are base-band signal,
fundamental frequency and harmonics only because the mixer is a direct conversion mixer.
The two-way configuration cancels the phase different of the output of the quadrature
coupler.
O90
50
The last section, IF filter, is simply a low pass filter with a large capacitor. It is easy to
extract the IF signal, base-band signal, because the closest unwanted frequency is 866MHz,
which is very large compared with the wanted signal, proximately 10kHz.
Figure 2.8.2 depicts the layout of the mixer. Strip-lines of the quadrature coupler are bent
in order to reduce the size.
Layout of the mixer
Figure 2.8.3
Figure 2.8.4 depicts the power of IF, which is DC, with different input power of LO and
RF. For a fixed LO power, 0dBm or 10dBm, the IF output power increases with RF-input
power. Nevertheless, if RF power is too low, the IF output stop dropping due to non-ideal
51
isolation between LO and RF; contrary, if the RF power is too high, there is a weird behavior
at the IF output because the effect of harmonics emerges.
IF (i.e. DC) output power with different LO and RF power
-60
-50
-40
-30
-20
-10
0
-40 -30 -20 -10 0 10
RF input power (dBm)
IF,
DC
(dB
m)
LO=0dBm
LO=-10dBm
Figure 2.8.4
The measurement result in Figure 2.8.4 stands on the ground of a DC output. Whatever,
it should also work in the case 10kHz base-band signal, the data rate of the RFID interrogator,
because 10kHz is in the pass-band of the IF filter, which is a low-pass filter.
52
9. A simplified interrogator
In order to test the data acquisition capability of the system, a simplified
interrogator is built. The interrogator contains five major functional blocks. The layout is
depicted in Figure 2.9.1
A Simplified Interrogator
Figure 2.9.1
The first functional block is a VTO, which gives the carrier signal. Most of the power
goes to the transmission path and some power couples to the mixer as the LO signal.
The second one is modulator. In order to make the result more obvious, a modulator with
100% modulation index is used.
The third one is a power amplifier.
53
The fourth one is the quadrature coupler, which imitate a circulator. The two output ports
are terminated with an almost-match load, which little amount of incoming signal is reflected.
The weak returned signal imitates the back-scattered signal from tags.
The last one is a mixer, which converts the RF signal back to base-band signal.
Due to non-ideal isolation of the quadrature coupler and incomplete matching of the two
“almost matched” loads, the RF signal goes to the mixer is sufficiently strong, therefore, LNA
is not required in the simplified interrogator.
Figure 2.9.2 depicts the experimental result of the transceiver
Modulating signal and Demodulated signal
Figure 2.9.2
The demodulated signal is weak and has much noise. One of the reasons is, although
individual functional block works properly, their measurements stand on the ground that both
source and load are Ω50 . However, when they are inter-connected, input and output
impedance of one functional block may affect that of other function blocks.
54
G. Cost Summary This project involves many RF and microwave discrete components.
RF transistors and diodes are bought from “www.rshongkong.com”, an electronic
products retailer. In fact, most of the components are cheap but buying each of them
individually is expensive. These components are about 60% cheaper if the purchase quantity
is more than 100. The price can be further reduced for a more bulky purchase.
Other materials, like dielectric substrate, SMT connectors, chip resistors, chip inductors
and chip capacitors can be obtained from microwave laboratory or supervisor.
55
Table 3.1.1 lists some of the major materials and components used in this project.
Table of expenditure
Materials and components Quantities Cost (HKD)
FR4 substrate ~ 4 pieces of A3 size $0 (from laboratory)
SMT connector 5 $0 (from supervisor)
RF transistor, BFR-183 15 $8.21
RF transistor, BFQ-19S 15 $10.20
PIN diode, BAR-64-05 8 $10.20
Tuning diode, BB-833 5 $5.84
Schottky Diode BAT-62 3 $15.70
Chip resistors, inductors and capacitors ~ 200 $0 (from laboratory)
solder --- $0 (from laboratory)
Others --- $100
Total cost $534.05
Table 3.1.1
56
Equipments used:
Microwave Network Analyzer
Microwave Signal Generator
Microwave Spectrum Analyzer
Cathode Ray Oscilloscope
DC Power Supply
Software used:
Agilent Advanced Design System 2004
IE3D
57
H. Discussions and Conclusions Interference
In the design of RFID interrogator for passive tags, a more complex design of the
interrogator trades off for the simple design of passive tags. For example, on one hand, tags
do not require generating carrier signal nor doing frequency translation, they can
communicate with the interrogator by back scattering. On the other hand, it is very difficult
for the interrogator to distinguish back-scattered signal from various types of interference.
In order to suppress the interference, hence, increase the detection range, the isolation
between the transmitting path and the receiving path is very important. In my design, I have
suggested a solution to due with this problem but how good is it is still a question.
There is a difficulty to simulate the system because it involves high carrier frequency and
base-band algorithm, therefore only transient analysis can be used but the computation power
required for the simulation will be extremely large.
An alternative is to assume the phase canceling loop works perfect and to use low
frequency signal to represent the envelope of the carrier frequency. I have tested the system
by this method in software simulator. It works as expected.
Design Technique
Software simulators play an important role through the whole project. Luckily,
58
simulators usually accurately predict the behavior of the circuit provide that the software
model of those components exist.
Measurement
When the RF power is large, say, above 20dBm, measurement becomes more difficult,
because the maximum power output of the microwave signal generator is 20dBm only. In
addition, RF cables become very lossy at high power. The loss can be as high as 5dB, which is
a disaster.
Other equipments, like network analyzer and spectrum analyzer, are easy to use and give
precious data.
What I have learned
I this project, I have learned many things. It strengthens my practical experience in
microwave circuit design.
Firstly, I learned how to start a project by following the international standard of a
product. Secondly, estimate system performance and design parameters for each functional
block set. Thirdly, simulate individual functional block by software simulator. Fourthly, carry
out measurements by various instruments. Finally, record design procedures and measurement
results. Also, writing project.
59
I. Reference [1] Guillermo Gonzalez, Microwave Transistor Amplifiers: Analysis and Design, second
edition
[2] http://www.daycounter.com/Calculators/Complete-RF-Amplifier-Design-Analysis-Calcula
tor.phtml
[3] http://ihome.cuhk.edu.hk/~s026680/MWcal.htm
[4] Constantine A. Balanis, .Antenna theory: analysis and design, 1938.
[5] David M. Pozar. Pozar, David M, Microwave engineering
[6] http://www.odyseus.nildram.co.uk/RFMicrowave_Circuits_Files/Attenuator.pdf
[7] Information technology automatic identification and data capture techniques — Radio
frequency identification for item management air interface — Part 6: Parameters for air
interface communications at 860-960 MHz, ISO/IEC FDIS 18000-6:2003(E)
[8] K. Kurokawa, “Some Basic Characteristics of Broadband Negative Resistance Oscillator
Circuits,” The Bell System Technical Journal, July 1969.
60
J. Appendices 1. Equation of maximum detection range
Let R be reader, T be tag, d be the distance between reader and tag, S be the received
signal power, n be the ratio of transmitting power coupled to LNA (interference). Also, P be
power and G be gain.
In forward transmission path,
Power density at Tag 24)(
dGPP RR
d ×××
=π
#A
Power received by Tag πλ
××
×=×=4
)(2
TdTdT
GPAPP #B
In backscattering path
Assume the tag backscatter all the power.
Power density at Reader 20 4)(
dGPP TT
×××
=π
#C
Power received by Reader πλ
××
×=×=4
)(2
00R
RGPAPS #D
Solving #A, #B, #C and #D, gives
4
4222
2
2
2 )4(4444 dGGPG
dGG
dGPS TRRRTTRR
×××××
=××
×××
×××
××××
=π
λπλ
ππλ
π #E
πλ
×××
×⎟⎠⎞
⎜⎝⎛=
4
41TRR GG
SPd #F
RPnS ×≥ is necessary in order to detect the backscattered signal. It is discussed in the
Challenge and Solution section, therefore,
πλ
×××
×⎟⎠⎞
⎜⎝⎛=
41)(
41TR GG
ndMax #1
61
2. Simulating result of 16 elements patch array antenna
Figure 4.2.1 shows the return loss, S11, of the 16-element patch antenna array. At the
desired frequency, the return loss is 17dB.
Return loss (S11)
Figure 4.2.1
62
Figure 4.2.2 shows the 3D diagram of the directivity of the 16-element patch antenna
array. The directivity is 16.6dBi
Directivity
Figure 4.2.2
63
Figure 4.3.3 shows the 3D diagram of the gain directivity of the 16-element patch
antenna array. The maximum gain is 11.5dBi
Gain
Figure 4.3.3
Efficiency of the antenna = 11.52 dB - 16.6 dB = -5.08 dB = 31 % #J
64
3. Stabilization of HBFP0405
Figure 4.3.1 shows the source and load stability circles on smith chart before
stabilization. Red circles show the source and load stability circles from 100 MHz to 5 GHz.
Circles in other colors show the gain of source and load conjugate matching. They overlap
with stability circles.
Source and Load stability circles before stabilization
Figure 4.3.1
The amplifier is conditionally stable and unilateral conjugate matching cannot be
applied.
65
Figure 3.3.2 shows the source and load stability circles on smith chart after stabilization.
The amplifier is still conditionally stable, fortunately, unilateral conjugate matching of source
and load at 866 MHz are located in the stable region for all the frequencies.
Source and Load stability circles after stabilization
Figure 4.3.2
66