20 views

Uploaded by write2arshad_m

RX Presentation

- FM Demodulation
- Gx2 Rx200 Install
- Harmony XB5R_XB5RFB01.pdf
- Linearity Enhancement Techniques _cmu Ece 2009 011 1
- RD500 Plastic Water Pipe Locator Rebrand
- Media Management and Entrepreneurship
- 7014166_Rev_A.pdf
- Gps Active Antenna Sl10913l
- E-Review EST Modules 3-4.pdf
- Two Key Questions About Using a Wireless Device in the Credit Card Industry Are
- eoi-902300
- Diff Between Correlation and Matched Receiver
- Simple Infrared Barrier
- Thesis_ Channel Multiplexer Design
- DRU Unit Description
- Anti Loss Device
- chang2017.pdf
- Reviewer 1
- FM1-130 Operation of the AN/USD-1 Surveillance Drone System
- Wireless Amplification

You are on page 1of 84

and measure them. Receivers are treated as black boxes with an input port at the antenna, and all what matters is the signal quality at a virtual output port at the detector input. Post-detector functionality is irrelevant at this stage. The useful definitions and results are brought in a format ready for use in design and analysis.

1

Not for distribution Intended only for students of the course Wireless Transceiver Design

Sensitivity (1)

The receiver (Rx) sensitivity, which we denote by Sens, is the smallest RF power at the antenna input and at operating frequency (on-channel, center-frequency), which will produce a useful signal at the input of the detector when no other RF signals are interfering at the antenna port, where useful means a signal that yields some predefined signal-power-to-noise-power ratio SNRd into the detector

Sens G = SNRd Nd

Definitions G = the center-frequency power gain from antenna to detector Nd = the total noise power at the input of the detector

2

Not for distribution Intended only for students of the course Wireless Transceiver Design

Sensitivity (2)

The noise power Nd is the result of integrating the thermal noise density (per unit frequency) due to all the circuits preceding the detector, over the total bandwidth seen at detector input. It turns out that a SNRd of 10 is a magic number, corresponding roughly to the level at which the receiver output starts becoming usable in the majority of common applications. In all that follows, we choose, as a rule of thumb, to define the receiver sensitivity as the RF signal level at the antenna port that will yield about 10dB SNRd. To this RF level corresponds a signal-power-to-noise-power ratio at the antenna input SNRi. The same SNRd figure for two different receivers may correspond to two different values of SNRi.

3

Not for distribution Intended only for students of the course Wireless Transceiver Design

Sensitivity is determined by: Noise figure: NF 10 log( SNRi / SNRd ) [dB ] RF channel bandwidth at detector input: B [Hz ] Minimal signal-to-noise required at detector SNRd [ - ] Once the above parameters are known, the sensitivity in dBm (dB above one mW) is computed by

Definitions Sens dBm = 10 log( Sens Watt / 103 ) = 10 log( Sens Watt ) + 30

log log10

4

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of sensitivity formula: The thermal noise power generated at antenna port over the bandwidth B seen at detector input is KTB [Watt], where K=1.381023 is the Boltzmann constant in MKS units, and T = 300 is the room temperature in degrees Kelvin. At antenna we have SNRi = Sens/KTB. Thus NF = 10log(SNRi/SNRd) = 10log([Sens/KTB]/SNRd). Computing Sens|dBm from NF above we get

Sens dBm = 10 log( Sens / 103 ) = 10 log( 10 NF / 10+3 KTB SNRd ) = NF + 30 + 10 log( KT ) + 10 log( B ) + 10 log( SNRd ) NF 174 + 10 log( B ) + 10 log( SNRd )

Where we used 10log(KT) = 10log(4.141021 )-204

5

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Consider a GSM cellphone with: - NF 8 dB - B 140 KHz - SNRd 10 Note: due to guard-band, B is somewhat smaller than the channel spacing, which is 200 KHz.

Sens dBm = 174 + 10 log( 140 103 ) + 8 + 10 log( 10 ) [dBm] 174 + 51.5 + 8 + 10 [dBm] = 104.5 [dBm]

2 / 50 v 1.3V @ 50 Sens Watt = 1010.45103 = vin in

6

Not for distribution Intended only for students of the course Wireless Transceiver Design

F1 = SNR | input #1 SNR | output #1 F2 = SNR | input #2 SNR | output #2

In #1

1st stage F1 , A1

Out #1

In #2

2nd stage F2

Out #2

F2

F = SNR | input #1 SNR | output #2

F2 1 F = F1 + A1

Power gain and Noise Figure dB Numerical Correspondence

Note: the Noise Figure of a passive element is equal to its attenuation

7

Not for distribution Intended only for students of the course Wireless Transceiver Design

BACKEND Device Gain (dB) Device NF (dB) 1st Device Gain (dB) Device NF (dB) Sens 0.43 V 1st LO BPF LNA 12 2 2nd 1st BPF mixer -1.5 -7 1.5 7 IF filter -3 3 IF amplifier 20 3.5 I - mixer 70 11 I LPF

-1.5 1.5

0o Phase o splitter 90

2nd LO

13. 7 -107.7

6.7 -114.7

3.7 -117.7

Q - mixer

Q - LPF

The NF at receiver input is computed step-by-step backwards by using the two-stage formula recursively, starting from the input of the backend, and going back towards the input of the 1st BPF . The overall noise figure, looking into each inter-stage point, is the ratio between the SNR at that point and SNRd (at detector input). The computed input noise figure NFi , yields the ratio between SNRi (the SNR at receiver input) and SNRd. Here we used our rule of thumb for sensitivity, i.e. SNRd = 10 and B=18 KHz.

8

Not for distribution Intended only for students of the course Wireless Transceiver Design

Notes

With the help of the computed NF and substituting the IF bandwidth for B , we are able to compute the power Sens of the RF signal required for sensitivity performance at the input of each stage in the receiver chain, with the previous stages taken out. The interim sensitivity and NF values obtained at the input of each stage, are useful during the design process, since they enable us to compare the measured interim values to the theoretical ones, confirming the accuracy of the design before we add the next stage. For computation purposes, the backend pass-band is taken the same as the IF pass-band, since both I and Q channels are 9KHz wide, yielding an equivalent noise bandwidth of 18KHz. In practice, in many receiver designs the overall IF bandwidth is made narrower as one proceeds towards the BB sampler, as this allows for building a Noise Squelch. One may find out that both sensitivity and noise figure, are worse at the input than near the backend. This fact does not imply that the front stages are badly designed, but is the outcome of the fact that sensitivity, in spite of being an easy spec to obtain, always comes at the expense of other crucial figures. In actual designs, sensitivity is frequently sacrificed along the receiver RF chain, in favor of intermodulation, spurious response etc. The subsystem values are taken from the SHR exemplary table.

9

Not for distribution Intended only for students of the course Wireless Transceiver Design

Backend Device Gain (dB) 70 Device NF (dB) 11 BPF Device Gain (dB) Device NF (dB) -1.5 1.5 LNA 12 2 Phase 0o o splitter 90 LO I - mixer I LPF

3.6 - 88.4

11 - 81

Q - mixer Q - LPF

Again we require SNRd = 10 , but here, we substitute for B the zero-IF bandwidth of 16MHz. The difference in sensitivity as compared to the SHR configuration is apparent, due to the difference in IF bandwidth.

What really matters is the difference in the input NF ( the right comparison to do ):

10

Not for distribution Intended only for students of the course Wireless Transceiver Design

The computation for VLIF sensitivity is essentially identical to the DCR case because: The line-up from antenna to backend input is identical The architectural differences in the backend itself do not affect the computations if the 3dB IF bandwidth is the same The shape of the IF filter itself is not a factor in the computation This will not be true with other specifications such as selectivity, image rejection etc.

11

Not for distribution Intended only for students of the course Wireless Transceiver Design

RF SIGNAL GENERATOR RECEIVER UNDER TEST

BER ANALYZER

BER Sensitivity

Set the RF generator level to minimum and adjust the frequency to the desired channel, then connect the generator output to Rx antenna port. Increase the RF level until the Rx output reaches the sensitivity performance, according to whatever figure is considered as usable (For instance, this may be a 10-2 BER value for digital equipment, a 12dB SINAD value for analog FM, or even the direct measurement of a SNRd value of 10 (or other required figure) if the detector input is accessible). The RF signal generator is modulated using a data stream generated by an instrument called BER analyzer. The BER analyzer collects the data stream recovered by the Rx detector output, compares it with the data stream fed to the RF generator and computes the BER.

12

Not for distribution Intended only for students of the course Wireless Transceiver Design

Block Sensitivity In packet data applications, for instance when using an IP protocol, it is customary to check the packet- error- rate

(PER). For small values of PER, the BER can be roughly estimated by dividing the PER by the packet size, since one may assume that, on the average, there is not more than one erroneous bit in each packet. In this case it is a good practice to use the shortest packet size available in order to reduce the chance of having more than one wrong bit per packet, which may lead to inconsistent results.

Analog FM Sensitivity

The RF signal generator is modulated using a 1000Hz sinusoidal audio signal. The BER analyzer is substituted with a SINAD meter. SINAD = (signal + noise + distortion) / (noise + distortion).

RF SIGNAL GENERATOR

AUDIO LOAD

SINAD METER

The SINAD meter places a notch filter on 1000Hz at the receiver audio output port, and computes the ratio between the total signal power (signal + noise + distortion) and the residual signal power (noise + distortion) after notch-out of the modulating sinusoid.

Setup Requirements

There are no special requirements on the RF signal generator performance (sensitivity is an easy measurements by means of equipment complexity). This will not be the case with most other specs.

13

Not for distribution Intended only for students of the course Wireless Transceiver Design

CCR is a measure of the SNRd at sensitivity level

Denote by Si = 2Sens the power of an on-channel RF signal which is twice the sensitivity power. Clearly, with Si applied at antenna port, the Rx output performance will be better than at sensitivity level. Denote by Scc the power of the smallest on-channel interfering RF signal that when applied to the antenna port simultaneously with Si, will cause the receiver output to worsen back to the performance obtained at sensitivity level. The co-channel rejection CCR, is the ratio, in dB, between the power values Scc and Sens

14

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of CCR

The computation of CCR is identical for all the architectures Proof of CCR formula with the interferer Scc and signal Si = 2Sens applied to antenna input simultaneously, the receiver performs as at sensitivity level, then SNRd is roughly the same as at sensitivity

Sens G = SNRd Nd Scc Scc 1 2 1 Sens + SNRd = SNRd Sens = SNRd Scc + Nd = 2 Sens Sens G SNRd

15

Not for distribution Intended only for students of the course Wireless Transceiver Design

INTERFERING RF SIGNAL (B)

COMBINING NETWORK

BER ANALYZER

First the interfering generator is shut off, and the level of the main generator (A) is increased until the output from the receiver fits the sensitivity performance (note that due to the attenuation of the combining network, the main signal level will be higher than sensitivity level). Then, the level of the main generator is increased by 3dB, causing the output performance (BER in this setup) to be better than the value at sensitivity. Now, the interfering RF signal generator (B) is modulated either with an analog or a digital signal, and its level is increased until the receiver output performance worsens back to the sensitivity value. The CCR is the ratio between the power level of the RF signal B and Sens. The attenuation of the combining network is assumed the same for both generators A and B.

16

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement using either BER, SINAD or PER is done using the corresponding setup. Care should be taken in the following: The modulation of signal B should be chosen so that it cannot be confused with the modulation of signal A. For instance, in analog FM measurements, if signal A is modulated with a 1000Hz sinusoid, then signal B will be modulated with a 400Hz sinusoid. The modulation level of signal B should be such that it does not cause the RF signal to exceed the channel bandwidth, otherwise part of the interfering power will be lost, and the measurement will be incorrect.

Example:

Assuming the magic number of 10 for SNRd the CCR estimate is

The actual figure measured in most standard two - way FM radios is around 7dB to 10dB.

17

Not for distribution Intended only for students of the course Wireless Transceiver Design

Selectivity (1)

Selectivity is a measure of immunity to close-in interferers

Selectivity is often used to mean Adjacent Channel Selectivity Consider again Si = 2Sens as defined for CCR. Denote the onchannel frequency by fR, and denote the spacing between adjacent channels by f. Denote by Sadj the smallest RF signal power at fR f. (one channel higher or lower that the operating channel) that, when applied to the antenna port simultaneously with Si, will cause the receiver output to worsen back to the performance obtained at sensitivity level. The Rx selectivity, which we denote by Sel, is the ratio, in dB between the power values Sadj and Sens, namely

18

Not for distribution Intended only for students of the course Wireless Transceiver Design

Selectivity (2)

As for CCR, the interference generated doubles the noise power at detector input, bringing back the SNRd to the same value as at sensitivity level. As opposed to the CCR case, the interference is due to a number of different mechanisms adding up at once: Rx synthesizer phase noise L [dBc/Hz] at fLO f. RF channel bandwidth B [Hz] at detector input. Synthesizer spurs SPRS [dBc] at fLO f. BB / IF filter rejection IFR [dB] at fIF f. As far as the measurement of selectivity is concerned, the phenomena above are indistinguishable, and all add up to the same effect of worsening the effective sensitivity measured in presence of a signal at adjacent channel frequency.

19

Not for distribution Intended only for students of the course Wireless Transceiver Design

Selectivity (3)

The limiting factors to selectivity are usually the synthesizer phase noise and spurs, not the IF filter shape and bandwidth (except in VLIF receivers).

The phenomenon is due to the fact that, even if the adjacent channel interferer itself is a perfectly clean delta function, the synthesizer phase noise and spurs phase-modulate it within the mixer (all the mixer is doing is summing/subtracting phases), thus causing the (mixed) spectrum to acquire the very same shape of the dirty LO, and to spread-out into the operating channel. Desired Signal Dirty Interferer Clean Interferer Ideal IF Filter Dirty LO

f

20

fIF - f

fIF

f0 + f

f0 + fIF

Not for distribution Intended only for students of the course Wireless Transceiver Design

f denotes an arbitrary frequency offset from LO carrier. The cumulative effect of the phenomena triggered by the presence of the adjacent-channel interfering power Sadj determines the selectivity Sel according to

Sel = 10 log( Sadj / Sens ) = 10 log [10 IFR / 10 + 10 SPRS / 10 + 10 SBN / 10 ] + CCR [dB ]

Where: SBN [dB] is the sideband noise power relative to LO power SPRS [dB] is the power of synthesizer spurs relative to Sadj IFR [dB] is the attenuation of IF or baseband filter at interferer offset

21

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of selectivity equation computing SBN

L(f) [dBc/Hz], denotes the SSB (single-sideband) phase noise, and is a property of the VCO & synthesizer combination. It decreases when increasing the absolute frequency offset |f | away from the carrier frequency and increases when increasing the LO frequency fLO. In essence this is a parasitic phase-modulation of the LO caused by thermal noise. By the mixer action, the LO phase-modulation is exactly transferred to the interferer, which spreads-out to the operating channel. The selectivity is related to the close-in (near fLO) value of the SSB noise power. Denoting by SLO the LO power , we have, at fLO + f

where we denote by PHNLO [Watt/Hz] the synthesizer phase noise power density at fLO + f, and dBc reads dB below carrier. We anticipate here that usually, for a given LO frequency fLO and for f in the range concerning selectivity, L(fLO ,f) behaves according to the simplified Leesons equation

L( f LO , f ) A( f LO ) 20 log10 ( f ) [dBc / Hz ]

For fLO fixed, we see that L(f) drops by 6dB each time the frequency offset away from the LO carrier is doubled. For very large offsets away from fLO the simplified equation is no longer valid, and L(f) reaches an ultimate floor. Note that [dBc/Hz] in is a convention rather than a dimension. It means that PHNLO at frequency fLO + f is

PHN LO = 10 S LO [Watt / Hz ]

22

Not for distribution Intended only for students of the course Wireless Transceiver Design

L 10

Proof of selectivity equation computing SBN (contd)

Denote by SBN (sideband noise) the value in dBc obtained integrating PHNLO / SLO over the channel bandwidth B. Denote by Sadjm = Sadj x G , the signal power resulting from mixing the adjacent channel interferer Sadj with the LO, and by PHNadjm the phase noise density of Sadjm at some offset f from its center frequency. Since Sadjm acquires the very same spectral shape of the LO, then the relative phase noise density PHNadjm / Sadjm as a function of the offset from Sadjm, is identical to the relative LO phase noise density PHNLO / SLO as a function of the offset from LO carrier frequency fLO, which produces at the detector input a relative power equal to SBN. If we consider PHNLO as being nearly flat near fLO + f , then

From the previous computation

and therefore

10 log( PHN LO / S LO ) = L

SBN L + 10 log( B ) [dBc]

23

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of selectivity equation computing SPRS and IFR

The synthesizer spurs are delta-type disturbances, and mostly depend on synthesizer hopping speed and loop filter construction. The most infamous are the reference spurs. The reference spurs modulate the interferer which spreads-out into the operating channel. Denote by Spm the power of the spur after mixing. If we have just one dominant spur, it is equivalent to an input signal of power Sp at adjacent channel, and the relative noise produced into the detector input is

The baseband / IF filter rejection IFR [dB] is the attenuation of the given IF or baseband filter at a function of the offset from carrier. We assume that IFR already reaches its ultimate value at fIF + f, and the attenuation of the interfering power reaching the detector due to Sadj is roughly IFR.

From the discussion above, and being G the center-frequency gain, the total interference power Ntot reaching the detector is

24

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of selectivity equation - conclusion

With the signal Si = 2Sens applied to antenna input, just as in the CCR case, the receiver performance is as at sensitivity level, then SNRd is roughly the same as at sensitivity, i.e.

[10 2 = SNRd

getting

IFR / 10

10

Sel 10

Taking 10 times the logarithm of the last equation completes the proof.

25

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

assume the following subsystem specifications : SSB noise at adjacent channel: L = -112dBc/Hz. IF attenuation within adjacent channel: IFR = 100dB. CCR = - 8dB. Channel bandwidth: B = 150KHz 10log(B)52 Synthesizer spurs: SPRS - 85dBc The selectivity computation yields

If we neglect the SSB noise (L = - ), we get Sel 77dB. This is a realistic example where the synthesizer phase noise is the limiting factor for the selectivity.

26

Not for distribution Intended only for students of the course Wireless Transceiver Design

In the SHR case, as in most narrowband receivers, the limiting factor for adjacent channel selectivity is the phase noise of the first LO.

The reasons are: The LO noise tail grows by about 6dB/octave when approaching fLO .The smaller the channel spacing, the closer the adjacent channel, and the closer the adjacent channel, the larger the noise tail power spread into the desired IF bandwidth. There is no protection against this interfering noise, since its frequency band falls exactly onto the operating channel. For a given value of f, the LO noise increases when fLO increases, thus the first LO is usually dominant as compared to the second LO, which has a much lower center frequency. In SHR receivers it is easy to obtain IF filters featuring extremely high rejection within adjacent channel. Moreover, there are reliable techniques to prevent occurrence of synthesizer spurs. Potential selectivity killers arising from parasitic interferences such as remodulation and microphonics, can also be kept low, especially in high-tier radios which are less cost-sensitive.

In most narrowband receivers, as far as selectivity is concerned, all the interferences except the LO phase noise may be neglected, getting

27

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Consider a SHR receiver with Channel spacing: 25KHz First LO SSB noise @ adjacent channel: L = -125dBc/Hz. CCR = - 10dB. Channel bandwidth: B = 18KHz. IFR = 100dB Using the simplified selectivity formula we get

BACKEND: BW = 18KHz (2 x 9 KHz) Front Filter 1 LNA Front Filter 2 1st Mixer IF filter IF BW = 18KHz Amplifier

28

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Computing the SSB noise from SHR selectivity If the LO characteristics are unknown, but it is known that the SSB noise is the limiting factor in determining the selectivity, then one may deduce the simplified Leesons equation by just measuring the adjacent channel selectivity value. Indeed, once the value of Sel at offset f is known, one may compute SBN using the simplified selectivity equation

Then the value of L at f is readily found from SBN and B

and thus the value of A(fLO) in the simplified Leesons equation can be computed

A( f LO ) L( f LO , f ) + 20 log10 ( f )

At this point, the simplified Leesons equation, allows the direct computation of L (and therefore of the selectivity) for any arbitrary offset f.

L( f LO , f ) A( f LO ) 20 log10 ( f ) [dBc / Hz ]

29

Not for distribution Intended only for students of the course Wireless Transceiver Design

In the DCR case too, the limiting factor for adjacent channel selectivity is usually the phase noise of the LO. DCR considerations When using the IQ architecture, the band-pass RF bandwidth seen at backend input is shared by the quadrature channels, each one of them being filtered by a low-pass filter half-bandwidth wide. Since in DCR there is no IF filter, the attenuation at adjacent channel depends entirely on the I and Q low-pass filters in the backend. In fully integrated architectures, usually the on-chip oscillator cannot provide a phase noise level low enough, thus one is unable to take full advantage of the attenuation of the on-chip filters, and the overall selectivity is poor even if the baseband low-pass filters have high attenuation. However, if one is willing to use an external VCO instead of the one built on-chip, good selectivity performance may be achieved.

30

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Assume the following values Channel spacing: 20MHz CCR = 10dB L = 135dBc/Hz @ 20MHz B = 16MHz IFR = 70dB Again we compute the selectivity using the approximated selectivity equation, getting

BACKEND: BW = 16 MHz (2 x 8 MHz) I LPF I - mixer BW = 8 MHz

Phase splitter

0o 90o

Q - mixer

Q LPF BW = 8 MHz

31

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

If the on-chip VCO is substituted by en external VCO in the previous example, a SSB noise level of 155 dBc/Hz @ 20MHz offset is pretty straightforward to achieve. In this case, the backend I and Q low-pass filters will become the limiting factor, and neglecting oscillator spurs, one can expect a selectivity value computed with the full selectivity formula

Specifically

In this case the backend selectivity becomes the limiting factor, and there is a 7dB improvement. To improve further, the backend low-pass filters must be made sharper.

32

Not for distribution Intended only for students of the course Wireless Transceiver Design

In the VLIF case, the limiting factor for adjacent channel selectivity is usually the complex polyphase filter performance VLIF considerations The case of hardware-based VLIF is different from the previous two. Due to the low-tier performance requirements, it is straightforward to conclude that the polyphase filter attenuation is the limiting factor for selectivity. The question is: how bad can we make the adjacent channel phase noise performance of the integrated VCO in this case? This is a proper question, since relaxation of SSB noise specs helps in reducing size, cost, complexity and current drain of the VCO, which may be critical in low-tier applications. Here the VCO noise and spurs may be neglected, if we assume that the polyphase filter attenuation is the dominant factor.

33

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Assume the following values Channel spacing: 20MHz CCR = 10dB B = 16MHz IFR = 35dB if we assume that the polyphase filter is dominant by one order of magnitude in the full selectivity equation, and neglect synthesizer spurs

and since we get the requirement

L SBN 10 log( B )

Such a noise performance (L) is very easy to obtain, and there is no use in looking for more. Computing the overall selectivity with CCR = -10dB, yields Sel 25dB . This figure is very poor, however, in many applications, this is all is needed (for instance in garage openers), while cost, size and current consumption of the receiver are critical.

34

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of Selectivity

The setup for the measurement of selectivity is identical to the setup for CCR

The measuring procedure is identical to the one used for CCR, except: The selectivity should be measured at both sides of the desired channel, namely one channel lower and one channel higher in frequency. The interfering RF signal generator (B) should be tuned to either adjacent channels, and the test repeated for both. The worst-case result is taken as the selectivity spec. There is no guarantee that the selectivity will be the same or even similar on both sides, especially due to asymmetric synthesizer spurs.

Unlike with CCR, selectivity measurements put severe constraints on the performance of the interfering signal generator B

Generator B must have a SSB phase noise at least 10dB better than the SSB phase noise of the synthesizer in the receiver under test. If not, one will measure a result worse that the actual Rx selectivity performance. If an adequate generator cannot be found, the only thing one can say is that the Rx selectivity figure is not worse than the measured one. The spurious level of generator B at adjacent channel must be at least 10dB below the selectivity figure to be measured, otherwise one gets again a wrong result. For generator A, as in the CCR case, there are no special requirements

Overlooking the performance of generator B is a common pitfall for RF engineers, and often yields puzzling measurement results.

35

Not for distribution Intended only for students of the course Wireless Transceiver Design

Intermodulation rejection is a measure of immunity to simultaneous multiple interferers

Intermodulation in the receiver context, usually refers to third - order intermodulation rejection, which is the prime intermodulation interference one should care about. Taking again Si = 2Sens and G the center-frequency gain as defined for CCR, consider two signals R2 and R4 both of the same RF power Sim, one of them centered at frequency fR + 2f (or fR 2f), and a second one of centered at fR + 4f (or fR 4f respectively), with fR being the received frequency and f the channel spacing, as defined for selectivity measurement. Set Sim at the smallest power value such that, when R2 and R4 are applied to the antenna port simultaneously with Si, the receiver output worsens back to the performance obtained at sensitivity level. The third - order intermodulation rejection, which we denote by IMR3, is the ratio, in dB, between the power values Sim and Sens, namely

36

Not for distribution Intended only for students of the course Wireless Transceiver Design

The third order intermodulation (IMR3) is due mainly to the fourth - order non - linearity of the mixer (s) in the Rx front end IMR3 considerations

IMR3 behavior is: increasing R2 and R4 by 1dB, namely, increasing the value Sim by 1dB, the interference generated at detector input increases by 3dB. Thus IMR3 is a third-order intermodulation. The phenomena is triggered by the presence, at the antenna port, of any couple of RF signals, Rn and R2n, one of them at frequency fR + nf (or fR nf) and the other at fR + 2nf (or fR 2nf respectively) , where n is an integer. This is because the fourth order nonlinear mixer (s) term

The

where fLO is the local oscillator frequency, produces, among the others, a signal proportional to A2BCcos(2fIF t), which is a product at IF frequency fIF that is indistinguishable form the wanted signal after mixing . If A and B are signals of equal amplitude, and since C is fixed, the spurious signal is proportional to A3. Thus the spurious signal grows faster than linear. The result is a noise signal N3 that adds up to the noise at the detector input.

37

Not for distribution Intended only for students of the course Wireless Transceiver Design

Spurious-generated noise power: Output Power [dBm] IPNo IPNi-p P Linear Product 1dB R 1dB (N1) x N dB 1dB PN G N-th order Product Nx x

PN = IPNo N( IPNi p )

IPNi-p x

p

38

IPNi

Not for distribution Intended only for students of the course Wireless Transceiver Design

The critical figure in determining the intermodulation is a receiver parameter called third order input intercept point, IP3i [dBm] Once IP3i is known, then IMR3 is computed using the formula

Conversely, once Sim has been measured according to the procedure described before, then IP3i can be readily computed substituting into the defining equation

Note If the radio front end provides some attenuation at the frequencies of Rn and R2n, then IP3i will be dependent on that attenuation, thus IP3i may become frequency dependent.

39

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of IMR3 equation With Si = 2Sens present, and Sim adjusted at a level such that the receiver performs as at sensitivity, as in the CCR case, the noise N3 must be roughly equal to Nd, thus, since

we get

Nd = 1 Sens G SNRd

Using dBm notation and recalling that

we get

40

Not for distribution Intended only for students of the course Wireless Transceiver Design

The value of IP3i is defined as follows: if we apply, at antenna input, a couple of interfering signals Rn and R2n of power IP3i each, they will generate, at detector input, the same power IP3o, in dBm, as if we

applied, at antenna input, a single on-channel signal of power IP3i. With G being the center-frequency gain from antenna to detector input, it follows from the definition of IP3i that

IP3o = IP3i + G dB

Intercept-point theory shows that a couple of interfering signals Rn and R2n of power Sim, generates at the detector input an interference N3 of power

Substituting IP3o and recalling that

and finally

41

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Assume the following specifications for a GSM phone Sens = 104.5dBm CCR = 8dB IP3i =-10dBm The computed IMR3 yields

Notes:

IP3i = -10dBm is the typical natural performance achieved using bipolar active mixers. Low Noise Amplifiers (LNA) and other RF circuits may also contribute to intermodulation, each one with a different IP3i value. If no one of them has dominant contribution, then the overall receiver IP3i will be some mathematical combination of the individual IP3i values.

42

Not for distribution Intended only for students of the course Wireless Transceiver Design

Minimize gain in front of distorting devices!

Every 1 dB of gain added in front of the non-linear device, causes the effective IP3i to degrade by 1 dB, yielding IMR3 degradation of 2/3 dB Indeed with a gain G1 in front of the the non-linear device we get

IMR3 = ( Sim dBm Sens dBm ) = ( [Sim G1 ] dBm [Sens G1 ] dBm ) = 2 ( IP3i [Sens G1 ] dBm ) + 1 CCR 3 3 = 2 ( [IP3i G1 dB ] Sens dBm ) + 1 CCR 3 3 = 2 ( IP3i Sens dBm ) + 1 CCR 2 ( G1 dB ) 3 3 3

IMR 3 without gain

43

Not for distribution Intended only for students of the course Wireless Transceiver Design

Effective IP 3i

RF SIGNAL (B) fR + nf ATTENUATOR >10dB MAIN RF SIGNAL (A) fR COMBINING NETWORK RF SIGNAL (C) fR + 2nf ATTENUATOR >10dB RECEIVER UNDER TEST

BER ANALYZER

First the interfering generators (B & C) are shut off, and the level of the main generator (A) is increased until the output from the receiver fits the sensitivity performance (note that due to the attenuation of the combining network, the main signal level will be higher than sensitivity level). Then, the level of the main generator is increased by 3dB, causing the output performance (BER or SINAD) to be better than the value at sensitivity. Next, the interfering RF signal generator (C) is FM modulated either with an analog or a digital signal, while the interfering RF signal generator (B) is left unmodulated. Now, the level of the interfering generators (B & C) is increased until the receiver output performance worsens back to the sensitivity value. The IMR3 is the ratio, in dB, between the (equal) power of generators B & C and Sens

44

Not for distribution Intended only for students of the course Wireless Transceiver Design

using either BER, SINAD or PER is done using the corresponding setup. The reason why only the farer signal should be modulated is that IMR3 involves the square power of the nearer signal, which would cause its FM modulation index to double, thus exceeding the channel bandwidth. Care should be taken in the following: The modulation of signal C should be chosen so that it cannot be confused with the modulation of signal A. For instance, in analog measurements, if signal A is modulated with a 1000Hz sinusoid, then signal C will be modulated with a 400Hz sinusoid. The modulation level of signal C should be such that it does not cause the RF signal to exceed the channel bandwidth, otherwise part of the interfering power will be lost, and the measurement will be incorrect. While there are no special requirements on generator A, IMR3 is a very demanding test as far as signal generators B and C are concerned.

Measurement

45

Not for distribution Intended only for students of the course Wireless Transceiver Design

Several points are worth an insight:

Some measurement methods are defined with n =1 (namely, the interfering signals are at one channel and two channels off). A quick glance to selectivity measurement procedure, reveals that with n = 1, the IMR3 result can never be better than the selectivity figure whenever the limiting factor is the synthesizer phase noise, or the phase noise of signal generator B (the one closer to the channel frequency fR). Thus, IMR3 measurements performed using n = 1 are somewhat meaningless. This is because usually the IMR3 figure is much better than the selectivity, so that one ends up measuring again the selectivity value. This is why it became customary to use n = 2, as described at the beginning of this section, but even then, the relief is not major, since the synthesizer (or generator) phase noise effect is reduced by only 6dB when n doubles . The best approach to measure the true IMR3 value, is to adopt the largest value of n, for which the signal at frequency fR + 2nf is still within the bandwidth of the RF front filters, and suffers no attenuation. The farer is the frequency fR + nf, the lesser the phase noise of signal generator and Rx synthesizer will affect the IMR3 measurement. Since the IMR3 figure is usually better than selectivity, even with relatively large values of n, still care must be taken in using low-noise signal generators for the interfering signals. The attenuators ( 10dB attenuation each) in the setup are an absolute requirement. This is because generators B and C usually are built so that, when delivering relatively high power levels they work without internal attenuator, by controlling the RF level to the driver of the output stage. Thus, unless external attenuators are used, one will end up measuring the IMR3 of the signal generators, which is usually much worse than the IMR3 of the receiver. This is a very common mistake that leads to wrong and puzzling results. In contrast, the on-channel generator (A) works at very low RF levels, near sensitivity, thus it is normally well-protected by its internal attenuator.

46

Not for distribution Intended only for students of the course Wireless Transceiver Design

Half-IF Rejection

Half-IF rejection concerns receivers with non-zero IF frequency. For DCR receivers, whose IF frequency is zero, there is an equivalent spec called second - order distortion rejection. Take again Si = 2Sens and G the center-frequency gain as defined for CCR, denote the center frequency by fR , and denote the IF frequency (including the VLIF case) by fIF. Consider a signal of power SIF/2 at frequency fIF/2 and the LO at fLO

f IF / 2 = f R 1 f , f LO = f R f IF 2 IF

where denotes upper-side of lower-side LO injection modes. Set SIF/2 at the smallest power value such that, when applied to the antenna port simultaneously with Si, the receiver output worsens back to the performance obtained at sensitivity level. The Rx half-IF rejection, which we denote by HIFR, is the ratio, in dB, between the power values SIF/2 and Sens

47

Not for distribution Intended only for students of the course Wireless Transceiver Design

HIFR considerations The HIFR behavior is: increasing SIF/2 by 1dB, the interference generated at detector input increases by 2dB. The half-IF spur is due mainly to the fourth - order non - linearity of the mixer (s) in the Rx front end. HIFR is a second - order intermodulation, since the disturbance grows (in dB) twice as fast as the interfering signal. The mechanism generating half-IF spurs is as follows: when the LO and half-IF frequency reach the mixer (s), the fourth - order nonlinear term

4 {A cos [2 ( f R 1 f )t ]+ B cos [2 f t ]} , LO 2 IF

f LO = f R f IF

produces, among the others, a spurious signal proportional to A2B2cos(2fIF t), which is at the IF frequency, and thus cannot be distinguished form the wanted signal after mixing. Since the LO amplitude B is fixed, the spurious is proportional to A2.

48

Not for distribution Intended only for students of the course Wireless Transceiver Design

The critical figure in determining the half-IF rejection is a receiver parameter called second - order input intercept point, IP2i [dBm] result is a noise signal, of power N2, that adds up to the noise at detector input. Once IP2i is known, then HIFR is computed using

The

Conversely, once SIF/2 has been measured, then IP2i can be readily

computed. If the radio front end provides some attenuation at the frequency fIF/2, then IP2i will be dependent on that attenuation, thus IP2i may become frequency dependent.

49

Not for distribution Intended only for students of the course Wireless Transceiver Design

Proof of HIFR equation With Si = 2Sens present, and SIF/2 adjusted at a level such that the receiver performs as at sensitivity, as in the CCR case, the noise power N2 must be roughly equal to Nd, thus since

we get

Nd = 1 Sens G SNRd

Using dBm notation and recalling that

we get

50

Not for distribution Intended only for students of the course Wireless Transceiver Design

The value of IP2i is defined as follows: if we apply, at antenna input, an interfering signal SIF/2 of power IP2i, it will generate, at detector input, the same power IP2o, in dBm, as if we applied, at antenna input, an on-channel signal of power IP2i.

With G being the center-frequency gain from antenna to detector input, it follows from the definition of IP2i that

IP 2o = IP 2i + G dB

By the definition of the second - order intermodulation phenomena, it follows that an interfering signal of power SIF/2, generates at the detector input an interference N2 of power

N 2 dBm = IP 2o 2( IP 2i S IF / 2 dBm )

Substituting IP2o and recalling that

and finally

51

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

Assume we measured the following for a GSM cellphone: Sens = -104.5dBm CCR = -8dB HIFR = 60dB Then we compute the second - order input intercept point

Notes:

The HIFR figure of (here 60dB ) is usually obtained with the help of a front filter attenuating signals at frequency fIF/2, which can be used only with superheterodyne. This filter is required since the natural IP2i achieved using bipolar mixers is around -7dBm, leading to about 45dB HIFR. The latter value is the performance in VLIF where no front filter can be added. Again, VLIF architectures are intrinsically low-performance.

52

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of HIFR

The setup for the measurement of HIFR is identical to the CCR case The measuring procedure is identical to the one used for CCR, except: The interfering RF signal generator (B) is tuned to

f IF / 2 = f R 1 f 2 IF

with the sign set according to whether the radio uses upper or lower side injection. In the case of superheterodyne, there are no special constraints on the interfering RF signal generator (B). In the Very-Low IF case, since the interfering signal is close to the channel frequency, all the restrictions on SSB noise and spurious required for the measurement of selectivity apply to RF signal generator B.

53

Not for distribution Intended only for students of the course Wireless Transceiver Design

Image rejection is not relevant for DCR whose IF frequency is zero. The image response of a receiver is a phenomena inherent to mixer (s) functionality Take again Si = 2Sens and G the center-frequency gain as defined for CCR.

Denote the center frequency by fR, and denote the IF frequency (including the Very-Low IF case) by fIF. Consider a signal of power SImage at frequency fImage

f Im age = f R 2 f IF

with the sign set according to whether the radio uses upper or lower side injection. Set SImage at the smallest power value such that, when applied to the antenna port simultaneously with Si, the receiver output worsens back to the performance obtained at sensitivity level. The image rejection, which we denote by IR, is the ratio, in dB, between the power values SImage and Sens

54

Not for distribution Intended only for students of the course Wireless Transceiver Design

The image spur is mainly due to the second - order non-linearity of the mixer (s) in the Rx front end. It is a first - order phenomenon, since the disturbance grows exactly as fast as the interfering signal.

The mechanism generating the image spur is the same which generates the desired mixing action: when the LO and image frequency reach the mixer (s), the second - order nonlinear term

produces, among others, a spurious signal proportional to ABcos(2fIF t), which is at IF frequency, and thus indistinguishable form the wanted signal after mixing. Since the LO amplitude B is fixed, the spurious signal is proportional to A. Being of first - order, the image spur bears no natural rejection at all. If we would measure it without additional protection at front-end level, we would come up with the same value as the CCR, namely, around 10dB.

55

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of IR (1)

IR considerations IR figures above 30dB are usually obtained with the help of a front filter attenuating signals at 2fIF offset, which can be used only with superheterodyne. In VLIF radios, where front filtering cannot provide image rejection, special architectures are used, which are able to provide an IR of about 35dB. Again we see that VLIF receivers are intrinsically low-performance. As far as the mixer (s) is concerned, a signal at frequency fImage is undistinguishable from an on-channel signal, and it results in a noise signal of power NImage that adds up to the noise at detector input. The only protection against it is the attenuation of a front-end filter or the rejection of a image-reject mixer, which we both denote by AImage. The image rejection IR is computed using

56

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of IR (2)

Proof of IR equation Again with G the center frequency gain from antenna to detector, and with SImage at the antenna port, the noise power NImage reaching the detector is

The noise NImage must be equal to Nd. Using

Nd = 1 Sens G SNRd

N Im age + Nd N Im age N Im age Nd 1 2 + 1 2 Sens G SNRd Sens G Sens G SNRd Sens G SNRd

Substituting for NImage in the rightmost expression, using and with dB notation

CCR = 10 log( SNRd )

S Im age AIm age 10 log( S Im age / Sens ) = IR = 10 log( AIm age ) + CCR Sens SNRd

57

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of IR

The setup for the measurement of image rejection is identical to the setup for CCR. The measuring procedure is identical to the one used for CCR, except: The interfering RF signal generator (B) is tuned to

f Im age = f R 2 f IF

with the sign set according to whether the radio uses upper or lower side injection. In the case of superheterodyne, there are no special constraints on the interfering RF signal generator (B). In the Very-Low IF case, since the interfering signal is close to the channel frequency, all the restrictions on SSB noise and spurious required for the measurement of selectivity apply to RF signal generator B.

58

Not for distribution Intended only for students of the course Wireless Transceiver Design

This specification is relevant for direct conversion receivers (DCR) whose IF frequency is zero. The phenomenon is triggered by the presence of a strong interferer anywhere nearby the center frequency, and produces baseband noise at the demodulator input. Since the IF of a DCR is at zero frequency, this is the analogous of the half-IF spur for heterodyne and VLIF receivers

Take again Si = 2Sens and G the center-frequency gain as defined for CCR. Denote

the center frequency by fR, and consider a signal of average power SD2 at some frequency fD2

f D 2 ( t ) = f R + f ( t )

where f(t) is any arbitrary low-frequency offset for which the signal is not attenuated by the filters in the RF chain. Set SD2 at the smallest power value such that, when applied to the antenna port simultaneously with Si, the receiver output worsens back to the performance obtained at sensitivity level. The second - order distortion rejection, which we denote by D2R, is the ratio, in dB, between the power values SD2 and Sens

59

Not for distribution Intended only for students of the course Wireless Transceiver Design

D2R considerations The second order distortion behavior: increasing SD2 by 1dB, the interference generated at detector input increases by 2dB. Thus, D2R is a second-order interference. The second - order distortion rejection of a receiver is caused mainly by the second - order non-linearity of the RF chain, namely LNA, mixer (s) and, in case of dual-conversion receivers, also amplifiers in the first IF chain. With a strong interferer SD2=A(t)cos(2fD2(t) t) present at antenna input, a second - order nonlinear term yields

where A(t) is a low-frequency signal modulating the amplitude of the interferer.

60

Not for distribution Intended only for students of the course Wireless Transceiver Design

D2R considerations

As opposed to the half-IF and third - order intermodulation cases, the LO is not involved with the generation of this spur. This is apparent from the fact that fD2 may have an arbitrary value within the front-end bandwidth. Since A(t) is low-frequency, we get a spurious signal proportional to the amplitude A2(t) which is at baseband frequency, and thus indistinguishable form the wanted signal after direct conversion mixing. Note that the spur bandwidth corresponds to the convolution of A(t) with itself. This yields a spectrum twice as wide as the bandwidth of A(t), and therefore the spectra of the spur may exceed the bandwidth of the baseband filter, making the definition of the D2R dependent on the bandwidth of A(t), not only on its power. Also, in defining D2R, the bandwidth of A(t) should not be taken too small since, in general, DCR architectures notch-out the DC frequency in the baseband, which may again result in an ambiguous definition of D2R. To avoid ambiguity, the bandwidth of A(t) may be taken half the baseband bandwidth.

61

Not for distribution Intended only for students of the course Wireless Transceiver Design

The critical figure, as in the half-IF rejection case, is the receiver parameter called second - order input intercept point, IP2i [dBm], and the result is a noise signal, of power N2, that adds up to the noise at detector input. Once IP2i is known, then D2R is computed using

The proof of is identical to the proof for HIFR replacing

Note: There is no front end protection for this spur, therefore, in a DCR the linearity requirements on the active RF components are much more severe than for heterodyne receivers, which reflects both in price and current consumption.

62

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of D2R

The setup for the measurement of second - order distortion rejection is identical to the setup for CCR. The measuring procedure is identical to the one used for CCR, except for: The interfering RF signal generator (B) is tuned somewhere within the front filter pass-band, and as far as possible from center frequency, so to avoid LO noise effects already seen in selectivity measurement. With this choice, there are no special constraints on the interfering RF signal generator (B). The interfering RF signal generator (B) should be AM-modulated with a modulating signal A(t) of bandwidth half of the receiver baseband bandwidth. Note: Since the D2R is a function of the interferer signal amplitude only, AMmodulation is all we need for measurement.

63

Not for distribution Intended only for students of the course Wireless Transceiver Design

Blocking (1)

The Rx blocking phenomenon is triggered by the presence of one or more strong interferers anywhere within the front-end pass-band, and away from channel frequency. again Si = 2Sens and G the center-frequency gain as defined for CCR. Denote the center frequency by fR, and consider a signal of power SBlock at any frequency fBlock

Take

f Block = f R + nf

where f is the channel width, and n is a large integer for which fBlock is still within the front-end pass-band (fBlock should be several channels away from fR, but not attenuated by the front-end filters). Set SBlock at the smallest power value such that, when applied to the antenna port simultaneously with Si, the receiver output worsens back to the performance obtained at sensitivity level. The Rx blocking, which we denote by Block, is the ratio, in dB, between the power values SBlock and Sens

64

Not for distribution Intended only for students of the course Wireless Transceiver Design

Blocking (2)

Blocking is one of the worst system killers in cellular and other

densely populated RF environments, and can be caused by unknown and alien nearby systems. Often it cannot be predicted or planned ahead of time, and may lead to a total system paralysis. Apart from a spurious-free and low-noise-floor synthesizer, there is no protection against it. The blocking is due to receiver LO far-out SSB noise-floor L() and synthesizer spurs interacting with an interferer away from channel frequency and producing a noise signal NBlock at demodulator input. Just as in the case of selectivity, even if the interferer consists of a perfectly clean delta function, the synthesizer phase noise and spurs modulate it within the mixer, causing its spectrum to acquire the very same spectral shape of the LO, and generating wide tails that spread-out towards the center frequency.

65

Not for distribution Intended only for students of the course Wireless Transceiver Design

If there are several strong interferers away from the channel

frequency, as it is often the case in a crowded RF environment, their tails sum up together and produce noise at demodulator input. For a single interferer, with SBN and SPRS as defined for selectivity the blocking is computed according to

The proof of is identical to the one for selectivity, setting IFR = and replacing and in the associated equations:

66

Not for distribution Intended only for students of the course Wireless Transceiver Design

Example:

It is of interest to compare the single-interferer blocking performance of two implementations of a GSM phone: The first using a standalone VCO with a noise floor L = -155dBc/Hz, The other using the system-on-chip VCO with a noise floor of L = -130dBc/Hz. Recalling that in GSM the IF bandwidth is B = 140kHz and , and assuming that spurs are negligible, we get

which yields

It is apparent that, basing the design on a system-on-chip, the built-in on-chip VCO cannot be used for high-performance radios.

67

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of Blocking

The

setup for the measurement of blocking is identical to the setup for CCR

The measuring procedure is identical to the one used for CCR, except: The blocking should be measured at a frequency fBlock several channels away from fR, but not attenuated by the front-end filters (usually n >10). If the far-out spectral picture of the LO is unknown, then the interfering signal B generating SBlock must be swept all over the front-end pass-band to catch the effects of unknown synthesizer spurs and noise bumps. Generator B must have far- out SSB phase noise and spurs at least 10dB better than the SSB phase noise of the synthesizer in the receiver under test, otherwise, one will measure a result worse that the actual blocking. Since the interferer is far from fR, this can be easily achieved by coupling the output of generator B through a band-pass filter.

68

Not for distribution Intended only for students of the course Wireless Transceiver Design

Dynamic Range

The Dynamic Range is a measure of the maximal gap between the sensitivity and the largest signal power that a receiver is capable to handle still delivering useful output

Receiver RF circuits cannot operate with indefinitely large input signals at

the antenna port. When a desired on-channel signal becomes too large, the receiver performance will begin to deteriorate, and at a certain input power level, it will worsen back to the sensitivity performance. Denote by Sat the largest RF power at the antenna input and at centerfrequency that, when applied to the antenna without other interfering RF signals, will cause the receiver output to worsen back to the performance obtained at sensitivity level. The Rx dynamic range DR is the ratio, in dB, between Sat and Sens

The dynamic range is mainly a result of a strong RF signal driving one or more circuits out of the proper operating/bias point, either by generating parasitic DC currents due to rectification of the RF power, which in turn change circuit bias, or by driving one or more output stages into saturation.

69

Not for distribution Intended only for students of the course Wireless Transceiver Design

There is no closed-form formula that fits every case, and the dynamic range is very dependent on receiver architecture and modulation type. For instance, in FM receivers, where deep circuit saturation is acceptable, there is virtually no limit to dynamic range. As opposed to that, in dense-constellation QAM modulation schemes, requiring highly linear receiver line-up, the dynamic range may be as low as 60dB, and adaptive attenuators at antenna port may be required for proper operation. Often the stages located closer to the detector are the more critical, since the gain from antenna port increases through the line-up towards the detector, and thus the signal level becomes higher making saturation more likely to occur. The measurement set-up for dynamic range is identical to the one used for sensitivity. The procedure is also the same, except that the measurement is done twice: First measure the sensitivity Sens. Then increase the input signal until the receiver output performance worsens back to the sensitivity value. The signal generator power at this stage is the value of Sat. Then compute

70

Not for distribution Intended only for students of the course Wireless Transceiver Design

Duplex Desense

In true-duplex radios (as opposed to TDD - Time Division Duplex), receiver and transmitter operate simultaneously at different frequencies, and the receiver input is isolated from the transmitter power by either a duplexer or in virtue of two separated Tx and Rx antennas. However, the Tx carrier tail at Rx frequency, due to synthesizer SSB phase noise, may physically reach the receiver input causing deterioration of the signal-to-noise ratio SNRd at detector. To restore the SNRd value required for sensitivity performance, the power of the desired signal must be increased, which results in an effective receiver sensitivity degradation. Let Sens be the receiver sensitivity measured with the transmitter disabled, and denote by TSens the receiver sensitivity measured with the transmitter operating simultaneously, which we refer to as duplex-sensitivity. The duplex desense DS is given by

71

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of DS (1)

Since

the Tx-Rx physical isolation is not perfect, some power from the transmitter tail, stretching-out to Rx center frequency, reaches the Rx input resulting in a noise power NDesense which adds-up at detector input. The mechanism is either the duplexer leakage, when one common antenna is used, or the direct electromagnetic coupling between Tx and Rx antennas, when separate antennas are used. We assume that the Tx carrier power is attenuated enough so, it does not cause receiver blocking. Assume: G : the power gain from Rx input to detector input. ATR : Tx- port output to Rx-port input attenuation at Rx center frequency. ST [Watt] :the transmitter power at Tx frequency. B [Hz] : the receiver bandwidth. L [dBc/Hz] : the SSB phase noise of the Tx synthesizer at Rx frequency.

Not for distribution Intended only for students of the course Wireless Transceiver Design

72

Computation of DS (2)

The phase noise PHNT generated by the transmitter at Rx center frequency is computed as for PHNLO in selectivity, substituting ST for SLO

PHNT = 10 ST [Watt / Hz ]

With the isolation ATR , the noise NDesense reaching the detector input is

L 10

It follows that, for sensitivity-level performance we need

L 10

Substituting

Nd = Sens G SNRd

one may compute DS exactly, however the computation is cumbersome. To get a feeling of the severity of the problem, it is enough to compute an approximate result instead

73

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of DS (3)

let us compute the required isolation if we allow a mild (3dB) sensitivity degradation. For this purpose we assume NDesense = Nd , and using the equation for NDesense

N Desense = Nd = 10 ST B G / ATR

which yields DS = 3dB

L 10

Then, substituting for Nd we get (note that here Tsens=2Sens)

L 1 = Nd = 2 10 ST B Sens = 10 10 ST B SNRd Sens G TSens ATR SNRd ATR L 10

Using the sideband noise value SBN =L+10log(B) as defined for selectivity, and the CCR, and taking the dB value finally yields

74

Not for distribution Intended only for students of the course Wireless Transceiver Design

Computation of DS (4)

Example:

in a typical scenario: ST = +27dBm Sens = -115dBm B =18KHz CCR = -8dB L = -150dBc/Hz at Rx frequency The required Tx-port to Rx-port isolation is

The above attenuation is readily attained using a duplexer, but very difficult to achieve with the use of two separate antennas. In a portable device, antenna-to-antenna isolation is of the order of 15dB. To achieve more, vertical pole-top physical separation is required.

75

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of DS

RF SIGNAL GENERATOR ATTENUATOR >30 dB Tx Rx Generator antenna antenna antenna RF SIGNAL GENERATOR TRANSCEIVER UNDER TEST TRANSCEIVER UNDER TEST

BER ANALYZER

BER ANALYZER

With Duplexer

Switch OFF the transmitter and record the generator level S1 (in dBm) that yields sensitivity performance using the procedure for measuring Sens. Switch ON the transmitter and record the generator level S2 (in dBm) that yields sensitivity performance using the procedure for measuring Sens. S2 is the duplex-sensitivity TSens. Compute DS as

DS dB = S 2 dBm S1 dBm

The attenuator is required in order to prevent the transmitter power from damaging the RF signal generator. Since we are interested only in the ratio TSens/Sens, there is no need to subtract the attenuator value from the level reading of the signal generator.

76

Not for distribution Intended only for students of the course Wireless Transceiver Design

Denote the receiver center frequency by fR, and the transmitter center frequency by fT . The duplex-image frequency fDI is the mirror of fR with respect to fT

f DI = fT + ( fT f R ) = 2 fT f R

Let TSens be the receiver duplex-sensitivity . Take TSi = 2TSens, and consider a signal of power SDI at frequency fDI . Set SDI at the smallest power value such that, when applied to the antenna port simultaneously with TSi, and while transmitting rated RF power at frequency fT, the receiver output worsens back to the performance obtained at duplex-sensitivity level. The duplex-image rejection, which we denote by DIR, is the ratio, in dB, between the power values SDI and TSens

77

Not for distribution Intended only for students of the course Wireless Transceiver Design

The duplex-image interference can be generated by several mechanisms. The most obvious, which we do not analyze here, is when, due to insufficient Tx-Rx isolation, the transmitter signal reaches the receiver input with insufficient attenuation, and acts as a parasitic LO signal. Under this condition, the LNA or the first mixer may act as an harmonic mixers. One less evident mechanism is when a signal at frequency fDI reaches the output of a Tx power amplifier (PA). Although the exact analysis of the phenomena is complex and strongly dependent on the characteristics of the PA, we can reach a qualitative understanding and a feeling of the order of magnitude of the interference by using a simple model for duplex operation. A very similar analysis holds for the case of separate antennas.

78

Not for distribution Intended only for students of the course Wireless Transceiver Design

Tx-Rx Antenna T Tx PA C DUPLEXER R RECEIVER

For the sake of simplicity, we assume that the output impedance of the PA exhibits the

behavior of a switch closing periodically at transmit frequency rate, either shorting to ground the duplexer terminal T, or leaving it disconnected. This is not a bad guess, especially with FM transmitters. From the above it follows that, if a signal at frequency DI = 2fDI coming from the common antenna terminal C reaches the PA output with voltage VDI , its instantaneous value

S ( t ) =

79

n =

[u( t nT ) u( t nT )] , T = 1 , 0 1 fT T

Not for distribution Intended only for students of the course Wireless Transceiver Design

Apart from an arbitrary time shift, and with T = 2fT, the nth harmonic component of S (t) has the form

sn ( t ) = ( 2 / n ) sin( n / T ) cos( nT t )

and the second harmonic component of S (t) has the form

s2 ( t ) = 1 sin( 2 ) cos( 2T t ) T

The amplitude of s2(t) reaches its maximal value of 1/ for / T = 1/4. Using the latter, the chopping product due to s2(t) is

With R = 2fR, it follows that DI = 2T R, and thus yields a term vR(t) at receiver frequency of the form

VDI vR ( t ) = cos( R t ) 2

80

Not for distribution Intended only for students of the course Wireless Transceiver Design

The above analysis shows that the natural protection of duplex-

image spur is of the order of 20log(2) 16dB. Any further protection should come from the attenuation ACT at frequency fDI in the path from terminal C to terminal T, and from the attenuation ATR at frequency fR in the path from terminal T to terminal R . Thus, using the very same approach as for image rejection we can roughly estimate a lower bound for the duplex-image rejection as

Usually the attenuation ATR at frequency fR, which is required for preventing duplex desense, provides enough protection for duplex-image.

81

Not for distribution Intended only for students of the course Wireless Transceiver Design

Measurement of DIR

INTERFERING RF SIGNAL (B) MAIN RF SIGNAL (A) COMBINER & ATTENUATOR (>30 dB)

BER ANALYZER

Switch OFF signal generator B. Switch ON the transmitter. Tune signal generator A to frequency fR, and measure the duplex-sensitivity TSens. Increase the level of generator A by 3dB (to a level equal 2TSens). Switch ON signal generator B, tune it to frequency fDI and increase its level SDI until the receiver output worsens back to sensitivity performance. Compute the duplex-image rejection

Notes: If DIR is better than blocking, the measurement will yield the blocking figure. The attenuator is required in order to prevent the transmitter power from damaging the RF signal generators. Since we are interested only in the ratio SDI /TSens, there is no need to subtract the attenuator value from the level reading of the signal generators.

82

Not for distribution Intended only for students of the course Wireless Transceiver Design

Half-Duplex Spur

Denote the receiver center frequency by fR, and the transmitter center frequency by fT. The half-duplex frequency fHD is located half way between fR and fT

f HD = 1 ( f + fR ) 2 T

The mechanisms causing the half-duplex spur are of higher order than the ones generating the duplex-image. This is because

f R = 2 f HD fT

which implies that a non-linearity of at least third-order must be involved. Therefore a natural protection of the order of 35dB or higher usually exists a-priori. Suitable mechanisms for generating the half-duplex spur are poor backward isolation of the final power device in the RF PA, or Tx power leakage to receiver input. We will not spend time analyzing here this mechanism, as it is usually of secondary effect. The measurement of half-duplex spur is carried out with a setup and procedure identical to the one described for duplex-image.

83

Not for distribution Intended only for students of the course Wireless Transceiver Design

Phantom-Duplex Spurs

When portion of the Tx power as small as 10dBm reaches the

receiver LNA or mixer, it can act as a phantom local oscillator signal producing IF frequency outputs whenever signal at the phantom image frequency fPH = fT fIF are present. Since the phantom image products are due to second-order non-linearity, no natural protection exists, as compared to prime mixing products. Just to get a feeling, for a 10Watt (+ 40dBm) transmitter, a Tx-Rx isolation of 50dB, although being enough to prevent receiver damage and saturation, will still generate considerable phantom spurs, unless additional attenuation for the phantom frequency is provided in the Rx front end. Along with the phantom image spur, the other spurs associated with regular mixer operation, such as phantom half-IF, will appear too. The measurement of the above spurs is done using the very same procedure used for the corresponding non-duplex figures.

84

Not for distribution Intended only for students of the course Wireless Transceiver Design

- FM DemodulationUploaded byRudra Pratap Singh
- Gx2 Rx200 InstallUploaded byStewie Griffin
- Harmony XB5R_XB5RFB01.pdfUploaded byZenon Nuñonca Zapata
- Linearity Enhancement Techniques _cmu Ece 2009 011 1Uploaded byA. Villa
- RD500 Plastic Water Pipe Locator RebrandUploaded byEnrique Murgia
- Media Management and EntrepreneurshipUploaded bygrinlia
- 7014166_Rev_A.pdfUploaded byPitoko Sauro
- Gps Active Antenna Sl10913lUploaded byStephanie
- E-Review EST Modules 3-4.pdfUploaded byleonardojlbc@yahoo.com.br
- Two Key Questions About Using a Wireless Device in the Credit Card Industry AreUploaded byShane Vashist
- eoi-902300Uploaded byjydee033
- Diff Between Correlation and Matched ReceiverUploaded bylibinpaul
- Simple Infrared BarrierUploaded byIonela
- Thesis_ Channel Multiplexer DesignUploaded byMoss OuaddaRi
- DRU Unit DescriptionUploaded byWelly Lukmanulhakim
- Anti Loss DeviceUploaded bydragon25
- chang2017.pdfUploaded byMohamed Bouhadda
- Reviewer 1Uploaded byMikhael Macha
- FM1-130 Operation of the AN/USD-1 Surveillance Drone SystemUploaded bydieudecafe
- Wireless AmplificationUploaded byandrewbaileykds
- Driveway AlarmUploaded bydivanshu16dec
- EE164 Project ReportUploaded byGurpartap Singh
- This is the final exam, 3rd year, communicationUploaded byhussein lafta
- E1XM Manual LowresUploaded bybroobenn3055
- Wireless Charging By RF SignalUploaded bygautam1992chaurasia
- 48279981 TV and FM Jammer SchematicUploaded byMihai Paun
- Mixers 1Uploaded byRamandeep Singh Grewal
- reportof AIR2Uploaded byshiva19892006
- vec820kUploaded byyu3za
- Tesla Nikola- the True Wireless -rare and prophetic newsclipUploaded byconfused1

- op-amp_driving_capacitance.pdfUploaded byjano2358j
- Evolution Carrier Aggregation InfoGraphicUploaded bywrite2arshad_m
- Zelda Isscc 2009Uploaded bywrite2arshad_m
- Decimal Powers of 10-Scientific Notation.pdfUploaded bywrite2arshad_m
- Decimal Powers of 10-Scientific NotationUploaded bywrite2arshad_m
- 5 Amplifier Design ADSUploaded bywrite2arshad_m
- Spru 889Uploaded bywrite2arshad_m
- Chapter 02 Basic Concepts in RF DesignUploaded bypriyanka patel
- 1MA98_12e_dB_or_not_dB.pdfUploaded byVenubabu Bandarupalli
- DCR TutorialUploaded bywrite2arshad_m
- j9Uploaded bywrite2arshad_m
- Huu SariUploaded bywrite2arshad_m
- why50ohmUploaded bywrite2arshad_m
- IP3NullingUploaded bywrite2arshad_m
- system design rfUploaded bywrite2arshad_m
- Reference Frequency StabilityUploaded bywrite2arshad_m
- Introduction 80211 Ac Technology and TestingUploaded bywrite2arshad_m
- Lesson 9Uploaded bywrite2arshad_m
- EuMW MicroApps Keyntote 2013 Steve CrippsUploaded bywrite2arshad_m
- ED_EE_III_2012Uploaded bywrite2arshad_m
- ReadmeUploaded byMuhyinDaud
- ReadmeUploaded byMuhyinDaud
- 29_09_2014_almazUploaded bywrite2arshad_m
- Lecture 04Uploaded bywrite2arshad_m
- RamadanUploaded bywrite2arshad_m
- Capacitor Tol SelectUploaded bywrite2arshad_m
- QFN_ANUploaded bywrite2arshad_m
- Love Chapter6Uploaded bywrite2arshad_m
- Love Chapter6Uploaded bywrite2arshad_m
- Behagi Turner Genesys Ch1Uploaded bywrite2arshad_m

- Unit 2 Satellite Comm LinkUploaded byK K SHARMA
- oldQAUploaded byjohn8880
- dB_-_Noise_2Uploaded byPatrick Belen
- Saturation Power.pdfUploaded byRitAux
- Max 2769Uploaded bymostafa
- poc2Uploaded bygheebiscuit
- NfaUploaded byamitshahom
- Chapter 1 - Introduction communicationsUploaded byGennie Brul
- NoiseUploaded byMuhammad Salah Elgabo
- Scanter 240100 TERMA Surface Movement Radar =System Specification.pdfUploaded byanteys
- ZX60-3018G+Uploaded byomarhanandeh
- 1652_MGA-86563Uploaded byRoman Zhuro
- 254bfb64c3779a0f4b0266a040dcca918586Uploaded byAhmed Zalook
- An106 - Low-noise Jfets - Spuerior Perform Ace to BipolarsUploaded byDewdeAlmighty
- Effect on Radar ReceiverUploaded byErdin Kamarudin
- IEEE - Fundamentals of RF DesignUploaded byAlex Alex Tudor
- Cisco Amplifiers 85102 Three Outputs 1GHzUploaded byMary V. Lopez
- Hf CommsUploaded byHimanshu Haldar
- ccsattema6-131230071303-phpapp02Uploaded byOussama
- Radar Range EquationUploaded byZahid Khan
- 2007 Holland Catalog EmailUploaded byfabianr8
- Macro-Micro Multi RRUs in One Cell(RAN16.0_Draft A).pdfUploaded bySam Ficher
- Spice-Simulation_Using_LTspice_Part_2.pdfUploaded byDaniel Lo
- Understanding Noise FigureUploaded byPraveen Singh
- The FundementalsUploaded byKide Bay
- 4. Amplification en 0Uploaded byABDUL GHAFOOR
- Doc Rf Mobistar2Uploaded byHatem MOKHTARI
- aisp cal reportUploaded bywebasf
- BiUploaded byraja_ramalingam07
- AV02-1283EN-DS-ATF-34143-08Jun20120Uploaded byEmilio Corti