The impact of interference on CDMA receiver

By Ryan Hendrickson will help determine the suscep-

Applications Engineer tibility to RF interference on
Agilent Technologies receiver test, as well as the pos-
Email: ryan_hendrickson@ sible solutions in resolving the
agilent.com interfering signals.
A mobile contains both a
When testing mobile phones, a transmitter (TX) and a receiver
primar y concern is how the (RX). The mobile TX emits
mobile will respond throughout levels up to +30dBM and as low
-20dB
a wide dynamic range. This is as -55dBm. The RX of the mo-
especially true with CDMA- bile will receive signals in the
based mobiles that can have as range of -20dBm to as low as
large as an 80dB dynamic range -108dBm. Both the TX and the
over a wide bandwidth, as com- RX of CDMA mobile phones
pared to standard analog or have a large dynamic range of
time-division multiple access over 80dB. The most critical
(TDMA) systems. This large levels to understand are the
dynamic range is a unique and maximum power (+23dBm on a Duplex
beneficial feature of CDMA class III mobile) and the RX sen-
technology. sitivity level (-104dBm).
CDMA mobile and base sta- In CDMA mobile testing,
tions are able to transmit lesser there are many tests that are
spectral power than other tech- performed to ensure that the
nologies. Since the mobile is mobile meets specific standard
capable of receiving signals be- requirements. The mobile’s TX RX TX
low -104dBm, noise levels power is tested to guarantee
within the given spectrum can that the range of power trans-
become influencing factors for mission is accurate over a wide Figure 2: Galvanic connections are the most commonly-used connection for mobile tests.
many of the tests outlined in the dynamic range. Specifically,
appropriate minimum perfor- the mobile’s maximum power to +30dBm for a class III mobile to accurately receive and de-
mance standards such as TIA/ transmission is tested to make phone. Since the power level modulate the forward link sig-
EIA-98C. It is important to certain that the mobile will that is measured by the instru- nal. Eb is the energy per bit of
understand these sources of transmit at a level that is close ment is large in comparison to the traffic channel and Nt is
potential interfering signals, as to the minimum of the specified noise contributions, noise is the total noise that is in the re-
well as the level at which they level for maximum effective usually an insignificant con- ceived bandwidth. This is simi-
are observed on the mobile’s isotropically radiated power tributor to the overall power lar to the concept of the analog
front-end. This information (EIRP). This range is +23dBm measured. The mobile’s TX is signal to noise ratio (S/N). As
also tested to meet minimum the Eb/Nt noise ratio is in-
transmission requirements. creased, the better the RX can
Specifications for the CDMA accurately demodulate the sig-
minimum power transmission nal. As Eb/Nt decreases, the
states that the mobile must more likely the mobile will in-
transmit less than -50dBm. accurately demodulate the for-
Even at this level, the influence ward link signal. The actual bit
of noise incident on the energy of the forward-link traf-
mobile’s transmission path is fic channel is 15.6dB below the
usually insignificant and will specification of total forward
not influence the measurement link power of -104dBm. In
of minimum power. Issues with other words, the actual sig-
measuring the minimum power nal that the mobile receives
of CDMA mobiles are usually for this measurement is
due to the noise floor limita- -119.6dBm. From this point
tions of the power meter used. on, we will refer to the forward
It is the sensitivity measure- link signal in the context of
Attenuation Duplex ment of the mobile’s RX that combined Walsh code power.
can be influenced by noise in- Depending on the margin
in dB cident on the mobile’s RX that has been designed into the
path. The specification for mobile’s RX, the sensitivity of
CDMA mobile RX sensitivity is the mobile’s RX to noise will
-104dBm. The mobile must be vary. It is typically a mobile
able to demodulate the for- with marginal performance
RX TX ward link signal that is being that will have issues with noise
transmitted at -104dBm with contributions on the forward
under 0.5 percent frame er- link path. There are many fac-
rors. Eb/Nt is the factor that tors that contribute to the over-
Figure 1: Cross coupling of the mobile’s transmitter to its receiver can contribute to noise. influences the mobile’s ability all Nt value in the Eb/Nt ratio.
Sources of noise break times, you will know Coupling of the mobile’s TX least-common test connection
KTB noise floor: The founda- where to start looking. to RX: Another source of noise method is directly through the
tion of the Nt value is thermal Mobiles tested in close pro- contribution to the overall Nt antenna of the mobile. We will
noise. This is the noise that is ximity: In a production envi- value is cross coupling of the assume that all discussed tests
always present in the environ- ronment, there are many test mobile’s TX to the mobile’s RX. are being performed through
ment. Thermal noise is also stations performing tests on This is an issue with the the physical “galvanic” connec-
known as the KTB noise floor, the mobile at various stages mobile’s design and is caused tor. Depending on the mobile
where: within the test plan. This when isolation between receive production process, the an-
K = Boltzmann’s constant means that one mobile being and transmission paths are not tenna will either be connected
(1.38 x 10-23) tested at a given stage could in- properly isolated or matched. or disconnected. In order to
T = Reference temperature terfere with another mobile Since this is a design issue, a look at the worst-case possibili-
(273 + °C) Kelvin testing at a different stage. Nor- remedy for this contribution is ties, we will assume that the an-
B = The effective RX noise mally, the mobile that is being costly and difficult to address. tenna is assembled to the mo-
bandwidth interfered with is being tested Since the forward and reverse bile phone.
for sensitivity. Recall that the links are separated by 45MHz In most cases, the mobile is
For CDMA systems operat- forward link is set to -104dBm. or 80MHz (for cell band and susceptible to noise through ei-
ing with 1.23MHz bandwidth, The primary interfering sce- PCS band, respectively), inter- ther poorly-shielded RF cables
thermal noise is about nario is when an adjacent mo- ference is highly isolated from or through the disconnected
-113dBm. You may be asking bile or mobiles are receiving a link to link. There can be in- antenna port on the mobile.
how the RX can demodulate a large forward link signal while stances of crosstalk that can be Typically, the antenna port can
-119dBm traffic level through a particular mobile is perform- noticed from one link to the be the primary source for un-
-113dBm noise floor. This is ing a sensitivity test with the other in the front-end of the desired noise, especially if the
due to the nearly 21dB of pro- for ward link signal set to - mobile. This is a factor of the antenna is connected during
cessing gain that CDMA has. 104dBm or below. The main mobile’s front-end design. testing. Figure 3 shows the
This is 14.4Kbps/9.60Kbps mobile tests that set the for- potential susceptibility of a
spread to a rate of 1.228Mcps ward link to a high level are Mobile RX path susceptibility mobile performing a sensitivity
(megacycles per second). mobile dynamic range, mini- The mobile phone is usually test to the forward link power of
Component noise: Noise mum transmit power and range tested through a physical RF -25dBm of another mobile.
from the RX’s front-end compo- of open loop. Typically, these connection. This connection is Mobile 1 represents many po-
nents (down converters and am- tests are performed at a forward most commonly referred to as a tential mobiles undergoing
plifiers) also contributes to the link level of -25dBm. “galvanic” connection. The testing with a forward link at a
Nt value that influences the sen-
sitivity of the mobile. This is the
extent to which the specified
sensitivity level of -104dBm is
-45dBm
meant to be the test level for the
forward link power. All other Mobile 1
noise contributions will raise Test instrument
-20dB
the noise floor while compro- -25dBm
mising the success of the sensi-
tivity test. A mobile with margin
designed in will have more room
for an increased noise floor as Duplexer
compared to a mobile with a Propagation loss:
marginal design.
Environmental noise: There 38dB of loss at a 1m separation of
are numerous unaccounted mobile 1 to mobile 2 RX TX
sources of noise that will de-
grade the Eb/Nt level of the for-
ward traffic channel. Any device
that radiates power can contrib-
ute to spectral noise. It will de-
pend on whether the signal itself -83.92dBm
or an intermod of two interfer-
ence signals fall within the Mobile 2
tested bandwidth at a significant Test instrument
power level determines whether -20dB
a problem will arise with sensi- -104dBm
tivity testing. Outside cell sites
can contribute to interfering
noise. This is especially true for
advanced mobile phone systems
Duplexer
(AMPS) when testing 800MHz
CDMA mobiles. Even micro-
wave ovens have been reported RX TX
as interfe-ring with mobile sen-
sitivity tests. In many produc- -103.92dBm = Sig
tion facilities lunch rooms may -110dBm = Noise
be close to the production lines.
If you see a large number of sen-
sitivity failures around noon or Figure 3: If the antenna is connected during testing, the antenna port can be the main source of undesired noise.
high power level of -25dBm. effects—any near-field contribu- still capable of receiving out- be quite effective in avoiding
Since mobile 1’s transmit path tions can be difficult to account side signals that can be super- interference. This requires
is at a level of -25dBm, the po- for); parallel alignment and no imposed on the TX and RX more sophistication in the
tential leakage through the dis- attenuating material between front-end. With the antenna control software to initializes
connected antenna port can be the two mobiles, the antenna of connected, it is also possible channel numbers and fre-
as high as -45dBm, assuming an mobile 2 can see a signal as high to transmit large signals into quencies, in order to elimi-
attenuation of the disconnect- as -83.92dBm. It is then pos- the environment while ma- nate frequency contention.
ing switch of 20dB. sible for the RX of mobile 2 to king measurements of the mo- • Frequency reuse can be real-
This attenuation value can experience an interfering signal biles TX at high power ranges. ized by paying close attention
var y depending on design. of -103.92dBm due to the 20dB By eliminating the antenna to the distance to which two
Since the signal leaking assumed attenuation of the an- ports match at the desired test identical frequencies are be-
through the antenna travels tenna to galvanic switch. This frequencies (removing the an- ing used. The mobile to mo-
through the air, it is subject to example shows the typical cause tenna from the antenna con- bile interference example
propagation attenuation that of a mobile producing undesired nector), the mobile will trans- may be useful in determining
can be characterized by using noise onto another mobile’s RX mit and receive signals at a a safe reuse distance.
Friis transformation equation. path. Due to the assumptions highly attenuated level.
made in this example, there are • Because environmental in- When dealing with RX inter-
λ2 many scenarios that could cause terference can find its way ference issues, especially in a
Pr = Pt G rG t
(4πd )2 varying results in actual imple- into the RF cabling of the test production environment, it
mentation. This example gives to mobile connection, it is al- may not always be easy to deter-
Where: a foundation for understanding ways important to use proper mine why yields are low due to
the path to which noise can be shielded cables. N-Type con- failed sensitivity tests. Loca-
r
P = Power received induced from one mobile to an- nectors with triple-shielded tion in the factory, time of day,
Pt = Power transmitted other. Orientation, distance, cables are most desired. test cell implementation, and
switch attenuation, shielding, • By shielding a mobile phone other seemingly irrelevant fac-
Gr = Antenna gain of TX and antenna design and imple- during testing, the environ- tors may seem to provide ran-
Gt = Antenna gain of RX mentation all play a role in the mental interference from dom failures of sensitivity. By
overall transfer of induced other mobiles or unknown stepping back and understand-
λ = Wave length power. spectral noise can be greatly ing the main sources of interfer-
d = Distance from TX and RX attenuated. The mobile- ing noise and how these sources
Points to remember to-mobile interference ex- of noise end up on the RX’s
Assuming: • If testing through the “gal- ample shown above may be front end, the task of resolving
vanic” connection, it is al- useful in determining the low yields in sensitivity testing
-9
Pt = 31.622 x 10 W ways better to test the mobile needed attenuation of shield- can be easily addressed. The
Gr and Gt = 0.09 with the antenna structure ing for your particular test steps regarding how interfe-
not connected. Since the de- needs. Shielding a mobile rence is imposed on a mobile’s
λ = 0.158m (1900MHz) sign of the mobile is around from outside interference is RX path, described in the mo-
d = 1m efficient transmission and re- not the only solution to elimi- bile RX path susceptibility sec-
ception of signals in the de- nating undesired noise. By tion, outlined the thought pro-
Assuming a worst case sce- sired band, if you are testing carefully managing the fre- cess taken in understanding
nario, mobile 1 is 1m from mo- through a physical Galvanic quencies at which all mobiles how interference will or will not
bile 2 (supposing only far-field connection, the antenna is are being tested on, you can effect RX sensitivity testing.