IP2 and IP3 Nonlinearity Specifications for 3G/WCDMA Receivers

Chris W. Liu, Morten Damgaard, Broadcom Corporation

The complete analysis of the nonlinearity requirements of a WCDMA direct conversion (Zero-IF) receiver is presented. This paper focuses on IP2 and IP3 requirements and covers all test cases defined in 3GPP, such as TX leakage, Intermodulation, Adjacent channel blocking, Narrow band blocking, In-band and Out of band blocking etc. The two tone IP2 and IP3 concept is further expanded into modulated signals. The formulas for computing IP2 and IP3 with modulated signals for different test cases are presented.

In a 3G/WCDMA mobile handset, the direct conversion receiver (DCR) architecture is commonly used due to its simplicity and low cost, where inter-stage filters are completely eliminated. However the linearity requirement for a direct conversion receiver is very critical. It is essential to define system linearity requirements properly for the receiver to meet the performance requirements. In FDD mode, the transmitter and receiver are continuously working at the same time. The transmitted signal leaks into the receiver due to limited TX to RX isolation of the duplexer and it is possibly the strongest interferer for the receiver in the handset and poses the most stringent linearity requirement for the receiver. I. 3GPP TEST CASES A number of test cases are specified in 3GPP for a WCDMA receiver, and each test case has different test conditions. Therefore nonlinearity performance requirements are needed of the receiver in each case. The related test cases are summarized below in Table A-F and are used through out the following discussion. A. Sensitivity requirement
Band I II III V VIII Test Conditions DPCH_Ec -117 -115 -114 -115 -114 TX power: 24 dBm <REFor> -106.7 -104.7 -103.7 -104.7 -103.7

E. Narrow band blocking (NBB)

Parameter DPCH_Ec or Iblocking Fuw offset Unit dBm/3.84 dBm/3.84 dBm MHz Test Conditions Band II, V Band III, VIII <REFSENS> + 10 dB <REFor> + 10 dB -57 -56 2.7 2.8 Tx power: 20 dBm

F. In band blocking (IBB)

Parameter DPCH_Ec or Iblocking Fuw offset Unit dBm/3.84 dBm/3.84 dBm MHz Test Conditions <REFSENS> + 3 dB <REFor> + 3 dB -56 10 Tx power: 20 dBm -44 15

The above test cases require the receiver to meet certain BER performance with these defined interferers. For RF system design, the requirements have to be interpreted into RF parameters such as Noise Figure (NF), Compression point (P1dB), Second Order Intercept Point (IP2), Third Order Intercept Point ( IP3), etc. II. SECOND ORDER NONLINEARITY The second order nonlinearity of the receiver will square the modulated blocker signal, such as the TX leakage signal, producing DC and low frequency components which fall into the receive band of the direct conversion receiver. The AM (amplitude modulated) signal is demodulated into the RX channel with twice bandwidth of the original interferers. Moreover, a strong blocking signal will also intermodulate due to second order nonlinearity with the TX leakage signal to create a TX image which can fall into the band. A. Mathematic Formula of IIP2 In general, the output signal of a nonlinear system can be described as follows:
V o = a 1 V i + a 2 V i2 + a 3 V i3 + K

B. Adjacent channel blocking

Parameter DPCH_Ec or Ioac Fuw Test Conditions Case 1 Case 2 <REFSENS> + 14 dB <REFSENS> + 41 dB <REFor> + 14 dB <REFor> + 41 dB -52 -25 +5 or -5 +5 or -5 Tx power: 20 dBm

C. Intermodulation (wideband and Narrowband)

Wideband Parameter DPCH_Ec or Iouw1 (CW) Iouw2 mean power (modulated) Fuw1 (offset) Fuw2 (offset) Test Conditions <REFSENS> + 3 dB <REFor> + 3 dB -46 -46 10 20 Tx power: 20 dBm Narrow band Band II, V Band III, VIII <REFSENS> + 10 dB <REFor> + 10 dB -44 -43 -44 -43 3.5 3.6 5.9 6.0

where Vo is the output and Vi is the input voltage. If using a traditional two tone signal as input Vi, we have: Vi = A cos(1t ) + A cos( 2 t ) The second order component of the output is:
Vo = a 2 Vi2 = a 2 A 2 (cos 2 (1t ) + cos 2 ( 2 t ) + 2cos(1t ) cos( 2 t )) = a 2 A 2 (1 + 1 1 cos(21t ) + cos(2 2 t ) + cos((1 + 2 )t ) + cos((1 2 )t ) 2 2

D. Out of band blocker (OOB)

Parameter DPCH_Ec or Iblocking (CW) Fuw (Band I) Unit dBm/3.84 dBm/3.84 dBm MHz Freq range 1 Freq range 2 <REFSENS>+3 <REFSENS>+3 dB dB <REFor> + 3 dB <REFor> + 3 dB -44 -30 2050<f <2095 2025 <f 2050 2185<f <2230 2230 f <2255 Tx power: 20 dBm Freq range 3 <REFSENS>+3 dB <REFor> + 3 dB -15 1< f 2025 2255f<12750

It shows that the second order IM2 products are created at three frequencies: DC, f1+f2 and f1-f2. In terms of power level, IM2 products are distributed against total IM2 power

as: 50% (-3 dB) at DC 25%(-6 dB) at f1+f2 25% (-6 dB) at f1-f2 The IM2 of low frequency is considered only in this paper since it is the one falling into band. The power level of the IM2 product at f1-f2 is 25% of the total IM2 power which is 6 dB below the total IM2 power. So the power level of IM2 at low frequency (f2-f1) can be expressed as:
Pim 2 ( dBm ) = 2 Pin IIP 2 6dB

the modulated downlink blocking signal is injected into the receiver with offset frequency at +/- 5MHz. By using method similar to the one discussed above, with an adjacent blocker signal (Test Model 1), the corrected formula for IIP2 with a WCDMA downlink signal is:

Pim 2 _ adj = 2 P2T IIP 2 3.87dB

(3)

(1)
dBm(S10)

100

B. IIP2 with WCDMA TX uplink leakage If the signal is AM modulated, such as TX leakage signal, the 2 tone formula above can not be fully applied. The difference between the formula and 2 tone signal is called correction factor [4]. The following section derives the correction factor for a WCDMA signal using ADS simulation. The ADS simulation Model is shown in Figure 1.

m2

m2 freq=1.000MHz dBm(S10)=-36.000

-100

-200

-300 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

freq, MHz

N_Tones N11

Figure 2 Low Frequency Output of Two tone simulation for IP2

DF DF
3GPP UL RF Source

3GPPFDD_RF_Uplink RF_Signal_Source

-20 -40

N_Tones N_CWblocker

SummerRF S3

FcChange F1

QAM_Demod Q1

dBm(S10)

-60 -80 -100

ResBW

SplitterRF S6 SplitterRF S5

SummerRF S7

SpectrumAnalyzerResBW S10

-120 -2 -1 0 1 2 3 4 5 6 7

MultiplierRF M1

GainRF G2

Figure 1 ADS Bench for IP2 Simulation

freq, MHz IM2_modulated_Power -45.722

The two tone test case and the case of single tone plus WCDMA uplink signal test case are simulated. The coefficients of the second order product in the model is set to 0.1 for simplicity since only the difference between using the two tone measurement and the TX uplink modulated signal is needed. The two tone simulation results are shown in Figure 2 and the simulation results with TX uplink signal are shown in Figure 3. With the TX uplink signal, since it is modulated, the low frequency IM2 products are measured by integrating the power from 1 KHz to 2.0MHz in the frequency domain. The difference of IM2 between the two tone and modulated signal is 9.72 dB. The IP2 two-tone formula for TX leakage case with correction factor is as follows:
Pim 2 _ txleakage = 2 Pin IIP2 15.722dB

Figure 3 Low frequency Output of WCDMA uplink simulation for IP2

D. IIP2 with Out of Band Blocker (OOB) In OOB test cases, depending on the blocker frequency, the IM2 products of the OOB blocking signal and TX leakage signal may fall into the RX band. The power of IM2 can be calculate using formula below.

Pim 2 _ OOB = PCW + Ptx IIP 2

III. THIRD ORDER NONLINEARITY

(4)

(2)

C. IIP2 with In-band blocking One of 3GPP test cases is adjacent blocker test, in which

For third order nonlinearity, the 3GPP intermodulation test case defines the IP3 requirements of the receiver. However the blockers such as adjacent channel blockers, narrow band blockers, out of band blockers, etc either leak into the RX channel, or cross-modulate, or intermodulate with the TX leakage signal, the distortion products falling

into wanted channel. The receiver needs to have good linearity performance under all blocker conditions.
dBm(Spectrum_Out_Overall)

A. IIP3 with Two Tone test Assuming the input signal as Vi with two tone signals, Vi = A1 cos(1t ) + A2 cos(2 t ) The output signal y(t) can be expressed as: 2 3 y (t ) = a1 Vi (t ) + a2 Vi (t ) + a3 Vi (t ) The third order intermodulation products are: y (t ) 3rdorder = 3a3 A12 A2 cos(1 ) 2 t cos(2 )t +
3a3 A1 A2 2 cos(1 )t cos(2 ) 2 t 3a A A2 3a A A1 cos(21 2 )t + 3 cos(22 1 )t + K 4 4 The third order IM3 products are created at: 2f1+f2, 2f1f2, 2f2-f1 and 2f2+f1. In a down conversion receiver, only the low frequency products are interesting. The power of IM3 at low frequency is expressed in following classical formula: =
2 3 1 2 2

-50

-100

-150

-200 1.92 1.93 1.94 1.95 1.96 1.97 1.98 1.99 2.00 2.01

freq, GHz

Figure 4 Two tone simulation

dBm(Spectrum_Out_Overall)

1) IM3 at 2f2-f1: PIIM 3 ( dBm ) = 2 P2 + P1 2 * IIP 3 2) IM3 at 2f1-f2 PIIM 3 ( dBm) = 2 P1 + P2 2 * IIP3 If two tones are equal, the formula becomes: PIIM 3 ( dBm ) = 3 Pin 2 * IIP 3

-50

(5)

-100

-150

(6)

-200 1.92 1.93 1.94 1.95 1.96 1.97 1.98 1.99 2.00 2.01

(7)

freq, GHz

B. IIP3 with Cross modulation The TX signal leakage signal can be cross modulated by a strong blocker such as a narrow band blocker, adjacent channel blocker, with the cross modulated signal falling into the receive channel. An example is illustrated in the Figure 4 and Figure 5. Figure 4 shows the output of a nonlinear circuit with two tone signal. Replacing one tone by a TX uplink leakage signal in Figure 5, cross modulation products are observed around the single tone at 1.99 GHz. For mathematical analysis, we could assume that the signal at the LNA input comprises of two interferers besides the wanted signal, namely a CW blocker signal and the TX leakage signal. Note that the TX leakage signal is amplitude modulated, while the blocker is not modulated in this analysis. Assuming the input signal x(t), with blocker and TX leakage signal then we get:
x ( t ) = A 1 cos( 1 t ) + A 2 [ 1 + m ( t )] cos(
tx

Figure 5 One tone is replaced by TX leakage signal

y (t ) = a1 x (t ) + a2 x 2 (t ) + a3 x 3 (t ) = a1 A1 cos(1t ) + a1 A2 (1 + m(t )) cos(1t )

2 + a2 A12 cos2 (1t ) + a2 A2 (1 + m(t )) 2 cos2 (tx t ) 2 + 2a2 (1 + m(t )) A1 A2 cos(1t ) cos(tx t ) + a3 A13 cos 3 (1t )

+ 2a3 A1 A2 cos2 (1t ) cos(tx t )

2

+ a3 A1 A2 (1 + m(t )) cos2 (1t ) cos(tx t )

2 2 + 3 a3 A1 A2 (1 + m(t )) 2 cos(1t ) cos2 (tx t ) 2 The term 3 a3 A1 A2 (1 + m(t ))2 cos(1t ) cos2 (txt ) can be

further expanded to:

2 3 / 2 a 3 A1 A2 (1 + m(t )) 2 cos(1t ) 2 + 3 / 2 a 3 A1 A2 (1 + m(t )) 2 cos(1t ) cos(2tx t )

The term,

2 3 / 2 a3 A1 A2 (1 + m(t ))2 cos(1t ) ,

shows the

t)

blocker signal being modulated by the square of the amplitude of the TX leakage. 1) Cross Modulation with adjacent channel blocker The straight forward thinking is to use the IIP3 formula with little modification. The quantity of the crossmodulation product can be evaluated with approximated formula:

where m(t) is the amplitude modulation having a fundamental frequency at WCDMA chip rate. When the signal inject into the non-linear RF circuit, receiver in this case, the output signal can be expressed as follows, considering only up to third order nonlinearity:

Pcrossmod = Cfactor + 2 PTX + Padj 2 IIP 3

(8)

where, Cfactor is the correction factor that takes into account the difference between using two CW tone measurement and modulated signal measurement. The correction factor has been determined by comparing the difference between two tone case and the TX leakage case in simulation. It is found that Cfactor is around 7.4 dB. The corrected formula for the adjacent channel blocker test case is: (9) Pcrossmod_adj = 2 PTX + Padj 2 IIP 3 7.4 2) Cross modulation with Narrow Band Blocker The narrow band blocker appears at +/-2.7(2.8) MHz, the offsets are much closer to the carrier compared the adjacent channel blocker case. The approximate formula is found to be: (10) Pcrossmod_nb = 2 PTX + Pnb 2 IIP3 2.4 C. Adjacent channel leakage - ACLR The receiver front end nonlinearity can create spectrum re-growth for an adjacent channel blocker, partially falling into the wanted band. The approximate formula to calculate the leakage power is reported in [3] as: (11) Paclr = 20.75 + 2 ( Pin IIP3) + 1.6 PAR where PAR is the Peak to Average ratio of the downlink signal. IV. SYSTEM SPECIFICATION In order to determine the IP2 and IP3 requirements, it is essential to consider all impairments including IM2, IM3 products, which must be low enough for the receiver sensitivity degradation to be acceptable. In the following sections, IP2 and IP3 are derived. The following analysis are applied to Band-I. A. IP2 at TX frequency requirement at Sensitivity level 1) Maximum allowed noise First of all, lets look at the maximum allowed noise floor without any interferers. From the 3GPP specification, the specified sensitivity is: Total input power: Ior = -106.7 dBm/3.84MHz, Dedicated physical channel power: DPCH_EC=Psens=-117 dBm/3.84 MHz. The processing gain of 12.2kps reference channel: Gp=25 dB It has been reported that the required Eb/Nt, the ratio of signal energy bit to noise spectra density, shall be better than 5.2dB to guarantee the Base Band (BB) modem can demodulate the WCDMA signal properly. In this paper, we assume Eb/Nt as 5.2 dB and add 2.2 dB implementation margin. Therefore the allowed total noise power at antenna input is: E Pn = Psens + G p b = 117 + 25 + 7.4 = 99.42dBm Nt Typically some margin is needed to cover process variation for production yield. By applying 1 dB margin, the total noise allowed at Antenna connector is:

Ntmax_ant=Pn-1=-100.4 dBm/3.84MHz At RFIC input, taking front end loss into account, the maximum allowed interfere level is: Ntmax= Ntmax_ant -Lfront=-104.2 dBm where Lfront=3.8 dB is assumed including switch loss and duplexer RX insertion loss etc. The distortion products due to component nonlinearity shall be kept below a certain level so that the receiver sensitivity will not be degraded too much. The interferers are not only from second and third order nonlinearity but also from other sources, which include: PA noise at RX band RFIC phase noise at RX band TX leakage IIP2 TX leakage reciprocal mixing Additionally, if blockers are present, other concerns also include: Blocker reciprocal mixing Cross modulation Adjacent blockerACLR Front end switch IM2/IM3 products 2) TX noise at RX band First all, the typical performance of the PA, RFIC and duplexer is studied in this case. Though the components from different manufacturers could perform differently, the typical performances used in this paper are summarized in Table 1, 2 and 3, based on major manufacturers datasheets.
Band 1 2 5 RXANT(dB)
Band

PA PA PA

TABLE 1 PA NOISE AND GAIN Gain 27 27 27

Rx Noise (dBm/Hz) -140 -140 -140

TABLE 2 DUPLEXER PERFORMANCE TX-ANT(dB) RX-TX (dB)

Insertion loss RX noise attenuation TX attenuation RX attenuation

Insertion loss

1 2 5

2.3 3.9 2.7

1.9 3.2 2.3

45 45 45

53 53 50

44 48 45

RFIC RFIC RFIC

TABLE 3 RFIC PHASE NOISE AT RX BAND Band Rx Noise (dBc/Hz) 1 -165 2 -165 5 -170

It should be noted that for the sensitivity test case, the transmitter output power is defined as maximum output power, 24 dBm, at the antenna connector. In all other test cases, the output power at the antenna connector is defined as 20 dBm according 3GPP specifications. The output power of the PA can be calculated by add loss

between the PA output and the antenna connector.

PAout = Pmax + FEloss + Duploss

where FEloss is the front end loss, including switch, PCB traces, matching loss. Duploss is the duplexer TX insertion loss. The RFIC output power can be calculated based on PAout and the PA gain. It is around 0.5 dBm, Prfic=0.5 dBm, in this case. The noise present in the RX band at the LNA input due to PA phase noise is Ppa_noise=PAnoise-ISOtx-rx+10*Log10(BW)=-118.15 dBm where ISOtx-rx is duplexer isolation from TX to RX at RX band, and channel bandwidth BW=3.84 MHz. And the TX noise at RX band due to RFIC is: Prfic-noise=Pn_rfic+PAgain+10*Log10(BW)-ISOtx-rx=-120.15dBm Since the total noise due to TX transmitter is the summation of both, it becomes:
Ptx _ noise = 10 log 10(10
PAtx _ noise 10

degradation is allowed, the total impairment power level must be 11.4 dB below the maximum allowed noise floor Ntmax. In general, it is good to keep IM2 due to TX leakage between 13 to 16 dB below Ntmax so that the IM2 product degrades the sensitivity by 0.1 to 0.3 dB only. In this paper 16 dB is chosen for better performance. If so, Pim2 shall be less than -120 dBm that is 16 dB below Ntmax. From equation (2), IIP2 at the TX frequency can be derived as:
IIP 2( dBm) = 2 P2T ( dBm) Pim 2 _ tx 15.722dB = 48dBm

Therefore the IIP2 requirement due to TX leakage for Band I at the RFIC input should be better than 48 dBm at the TX frequency. With IIP2=48 dBm, adding all other impairments, the overall sensitivity degradation is less than 0.6 dB. Further improvements in the PA phase noise and duplexer attenuation will continue to minimize the degradation. B. Adjacent channel blocker case 1 The adjacent blocker appears at 5 MHz offset with a power level of -52 dBm at antenna connector. Since the input wanted signal is 14 dB higher than sensitivity, it is reasonable to assume that the degradation due to the total interferer power shall be less than 14 dB. However in practice, margin is needed. Note that with an adjacent channel blocker, the major contributor of in channel distortion is from ACLR, rather than IP2. First of all, we need to determine the power leakage into the channel due to adjacent channel blocker, Paclr. To minimize the impact of ACLR, in this paper, Paclr is kept 16 dB below the wanted signal. The allowed total noise floor Nt, is 14 dB higher than Ntmax in this test case, therefore Nt = -90.2 dBm, The allowed Paclr will be Paclr= Nt -16=-106.2 dBm. From the equation (10), IIP3 can be calculated,
IIP3 = 1 ( Paclr 20.75 + 2 Padj + 1.6 PAR ) = 7.2dBm 2

+ 10

Pr fic _ noise 10

) = 114.86dBm

3) TX leakage reciprocal mixing The TX leakage power at the RFIC input is:

Pleakage = PAout ISOtx _ rx = 25.6dBm

If assuming the phase noise of the RX LO at the TX frequency is -157 dBc/Hz, then the reciprocal mixing at the RX band is: Ptx_reciprocal=-116.75 dBm 4) TX Leakage IIP2 First, lets make an assumption that the allowed sensitivity degradation as X dB, and then we will have:

10

Ptx _ noise Ptx _ im 2 Nt max Ptx _ reciprocal + + + ) 10 10 10 10

= 10

Nt max + X 10

We also need to check the IP3 requirements due to Cross modulation. Based on the cross modulation formula derived in previous section, equation (8), the cross-modulation product can be calculated:

Arranging the equation above and expressing the above equation in dB, we get:

Pcrossmod = 2 PTX + Padj 2 IIP 3 7.4 = 108.27dBm

The cross modulation product with IP3=-7.2 dB is 18 dB below the Nt, not a problem. Then Pim2_adj due to adjacent channel blocker can be assumed 16 dB below the Nt as well. Then IIP2 at 5MHz adjacent channel, IIP2_adj is obtained:
IIP 2( dBm) = PIIM 2 _ adj + 2 Padj 3.87dB = 15dBm

10 log 10(10

Ptx _ noise 10

+ 10
X

Ptx _ im 2 10

+ 10

Ptx _ reciprocal 10

= N t max + 10 log 10(10 10 1)

X

The term 10 log 10(1010 1) is the factor that determines how much the impairments shall be below Ntmax. For example if X=0.3 dB, meaning 0.3 dB sensitivity

C. Adjacent channel case 2 The blocker is present at 5 MHz offset with a power level of -25 dBm and the input wanted signal is 41 dB higher than sensitivity. In this case, Nt= -63.2 dBm, the allowed Paclr=Nt-16=-79.2 dBm. The IIP3 and cross modulation product are calculated below.
IIP 3 = 1 ( Paclr 20.75 + 2 Padj + 1.6 PAR ) = 6.2dBm 2

Pcrossmod = 2 PTX + Padj 2 IIP 3 9.67 = 110.7dBm

Similar to the adjacent channel blocker case 1, IIP2 can be derived for the adjacent channel blocking case 2:

With the OOB test case, not all blocker frequencies produce distortion that falls into wanted band. For the second order nonlinearity, we will consider: Fcw=Frx-Ftx Fcw =Frx+Ftx For the third order nonlinearity, we will consider: Fcw = Frx -F/2 Fcw = Frx -2*F Fcw =2* Ftx + Frx Fcw =F /2 where Fcw is the blocker frequency, Frx is the RX channel frequency, Ftx is the TX frequency related to Frx, and F is the frequency separation between Frx and Ftx. 1) Frx-Ftx The CW blocking signal is at very low frequency, which is the separation between RX and TX frequencies. The second order IM2 results in mixing of Ftx+ (Frx-Ftx) = Frx. As the CW blocker level is -15 dBm, and the duplexer attenuation at the CW blocker frequency is around 30 dB for Band I, we get Pcw=-45 dBm at the input of the RFIC and the TX leakage power level Ptx=-29.6 dBm. Using equation (4), with Pim2 16 dB below Nt, which is 3 dB above noise floor of -104.2 dBm, IIP2 is calculated as:

IIP 2( dBm ) = PIIM 2 _ adj + 2 Padj 3.87dB = 12dBm

D. In band blocker 10MHz and 15 MHz The input wanted signal is 3 dB higher than sensitivity and the blocker at 10 MHz offset has a power level at -56 dBm, and -44 dBm at 15 MHz offset. Since the offset frequency is far from the carrier, there is no concern for ACLR in this case. Using equation (3), keeping the IM2 product 16 dB below the Nt, IIP2_adj at 10 MHz and 15 MHz are computed:
IIP 210 MHz = PIIM 2 _10 mhz + 2 P10 mhz 3.87dB = 12dBm

IIP 2 = PCW + PTX PIM 2 = 38dBm

2) Ftx+Frx In this case, the OOB signal is at high frequency. The mixing of (Ftx+Frx)-Ftx=Frx, falling in the RX band. The blocker level is -15 dBm, and the duplexer attenuation at the blocker frequency is around 30 dB for Band I so that Pcw=-45 dBm, and Ptx=-29.6 dBm. Using equation (3), with Pim2 16 dB below Nt, we have

IIP215MHz = PIIM 2 _ 15mhz + 2 P15mhz 3.87dB = 12dBm

E. Intermodulation For band I, the narrow band blocking test case does not apply, but for other bands. It does in the wideband intermodulation test case, one of the blocking signals is a CW tone at -46 dBm, while another blocker is a WCDMA modulated signal with power level of -46 dBm. The wanted signal is 3dB higher than sensitivity power level. Keeping the IM3 product 16 dB below the Nt which is 3dB above Ntmax we can compute the IIP3 using equation (7). IIP 3 = 0.5 * (3 Pin PIIM 3 ) = 15dBm

IIP 2 = PCW + PTX PIM 2 = 38dBm

3) Frx -F /2 The blocker power level actually varies depending on frequency offset according the 3GPP requirements. When Fcw is between 2050MHz and 2075 MHz, the blocker level is -44dBm. As the duplexer attenuation at blocker frequency is around 5 dB, Pcw is -49 dBm. IIP3 is calculated as:
IIP 3 = 0.5( 2 PCW + PTX Pim 3 ) = 7dBm

F. Out of band blocker For Out of band OOB blocking cases, we need to consider both the second order and third order distortion from the switches that is between the duplexer and the antenna. The switches in the front end module generate second and third order nonlinearity components falling in RX band as well. Bases on manufacturers datasheet, it is assumed in this paper that IM2 and IM3 of the front end switch module are: Psw_im2=-110 dBm Psw_im3=-110 dBm

When Fcw is located between 2025MHz to 2050 MHz, the blocker power level is -30dBm. Assuming duplexer attenuation at blocker frequency is 15 dB, Pcw is -45 dBm. IIP3 is calculated as:

IIP 3 = 0.5( 2 PCW + PTX Pim 3 ) = 3dBm

When Fcw is between 2015MHz and 2015 MHz, the

blocker power level is -15dBm. With the duplexer attenuation at the blocker frequency of 30 dB, Pcw is -45 dBm. IIP3 is calculated as:
IIP3 (dBm)

IP3 Wide band intermodulation

0.00 -5.00 -10.00 -15.00 -20.00 2110 IP3_IMD, positive

IIP 3 = 0.5( 2 * PCW + PTX Pim 3 ) = 3dBm

4) Frx-2*F The CW blocker frequency is located between 1730MHz and 1780 MHz. To allow Pim3=-117 dBm, which is 16 dB below the max allowed noise floor Nt, the blocker power level is -15dBm. Assuming duplexer attenuation at the blocker frequency is 38 dB, Pcw is -54 dBm. IIP3 is calculated as:

2120

2130

2140

2150

2160

2170

Frequency (MHz)

Figure 7 Measured Results of IIP3 Wideband Intermodulation

IIP 3 = 0.5( PCW + 2 PTX Pim 3 ) = 1dBm

5) Fw=2*Ftx+Frx When the blocker frequency is between 5950MHz and 6130 MHz, the CW blocker has a level of -15 dBm. As the duplexer attenuation at the blocker frequency is around 30 dB, Pcw is -45 dBm. With Pim3=-117 dBm, which is 16 dB below max allowed noise floor Nt, IIP3 is calculated as:

VI. CONCLUSION The system analysis of nonlinearity of a WCDMA receiver is discussed. The formulas for IP2 and IP3 for all test cases are presented and applied. The formulas are useful for system analysis. The assumed margins can be different in each test case. The formulas may be slightly different from other published results due to difference in simulation setup though. ACKNOWLEDGMENT The authors would like to thank colleagues at Broadcom Corporation for support, discussion and review. REFERENCES
[1] [2] [3] http://www.3gpp.org http://www.broadcom.com Harald Pretl1, Linus Maurer1, Werner Schelmbauer1, Robert Weigel, LINEARITY CONSIDERATIONS OF W-CDMA FRONT-ENDS FOR UMTS, Microwave Symposium Digest., 2000 IEEE MTT-SVolume 1, 11-16 June 2000 Page(s):433 - 436 Walid Y. Ali-Ahmad, Effective IM2 estimation for two-tone and. WCDMA modulated blockers in zero-IF, in RFDesign, Apr 1, 2004 Ranta, T.; Ella, J.; Pohjonen, H., Antenna switch linearity requirements for GSM/WCDMA mobile phone front-ends, Wireless Technology, 2005. The European Conference on 3-4 Oct. 2005 Page(s):23 - 26 Qizheng Gu, RF System Design of Transceivers for Wireless Communications,Springer.

IIP 3 = 0.5( PCW + 2 PTX Pim 3 ) = 5dBm

6) Fw=F /2 In this case the blocker is defined as -15 dBm CW signal, and the duplexer loss at this frequency is 30 dBm. Then Pcw is -45 dBm. Similar to the previous case, to have Pim3 16 dB below max allowed noise floor Nt, IIP3 is calculated as:
IIP 3 = 0.5( 2 PCW + PTX Pim 3 ) = 5dBm

V.

MEASRURED RESULTS

[4] [5]

The IIP2 and IIP3 have been fully verified on a Broadcom WCDMA transceiver IC. The results of IIP2 at TX frequency and IIP3 wideband intermodulation are shown in Figure 6 and Figure 7, respectively. Both results show that the receiver has excellent linearity.
IIP2@TX frequency
60 55 50 IIP2 (dBm) 45 40 35 30 25 20 2110 2120 2130 2140 Frequency 2150 2160 2170 IIP2@TX

[6]

Chris W. Liu received his M.Sc from University of Montreal, Canada and B.Sc from Tianjin University, China. Chris has more than 17 years experience in the wireless communications. He is currently with Broadcom as Principal Systems Engineer. Before joining Broadcom, he holds various positions in RFMD, Intel, Focus microwaves etc. Chris can be reached by email at: chrisl@broadcom.com. Morten Damgaard received his M.Sc. degree from University of Aalborg, Denmark in 1989. Morten has 20 years of experience in the wireless and cellular industry doing cell phone RF design, semiconductor RF system design, and radio architecture design for GSM, EDGE, WCDMA, and multi-mode transceivers and radios. He currently works with Broadcom Corporation in Irvine, California as a Senior Manager of Cellular RF System Engineering. Before joining Broadcom Morten has worked for Axiom Microdevices, Skyworks Solutions, Conexant, Rockwell Semiconductors, and Dancall Telecom. Morten holds 31 U.S. patents, mainly in the field of cellular radio design.

Figure 6 Measured Results of IIP2 at TX frequency