Tropper Technologies Homepage

 

Wireless Workshop - Explore Digital Modulation Circuit Impairments

Tropper Technologies' Wireless Workshop is designed to explore various aspects of emerging wireless technologies...




As almost any modulation scheme can be realized by using I/Q modulators, they are very useful and have become extremely popular in today's digital wireless transceivers.  There are several factors that can affect the behavior of the I/Q modulator.  These impairments in the I/Q modulators directly influence the performance of the systems in which the transceivers are employed.   Here we will focus specifically on the effect of transceiver impairments in the wireless LANs and on the IEEE 802.15 Bluetooth standards.
 
  • I/Q Modulators

                                     

  • Digital Modulation Impairments

Digital modulation impairments can be classified as either linear impairments or non-linear impairments.  Examples of linear impairments are I/Q amplitude imbalance, phase imbalance and I/Q DC offset.   Examples of  non-linear digital modulation impairments are LO phase noise and the Power Amplifier non-linearities associated with amplification of the I/Q modulator's output signal.  Any of the above I/Q modulator impairments can cause errors elsewhere.   By way of example, phase imbalance in the I/Q modulator can cause unbalanced transmission through the two sides of the modulator.  When the individual vectors are summed in the combiner, undesired sidebands might be produced.  In addition, the symbol constellation states will be skewed, resulting in sub-optimal symbol detection in the receiver.   Similarly, if a PLL is used to synthesize the LO frequency, then phase noise introduced by the PLL (due to the divider circuitry as well as VCO noise) can degrade the Rx sensitivity.

 

  • General Transceiver Performance Metrics

The various digital modulation impairments mentioned above can all degrade the performance of the wireless transceiver.  There are several commonly used metrics to gauge the component performance of various aspects of wireless transceivers:

  • Noise Figure

Noise is a random process, present even in the absence of a signal.  When "white" noise passes through frequency selective components (like filters), it becomes "colored" noise.  In any case, noise can be modeled by various distribution functions.

Noise Figure (dB) = A measure of the amount of noise added by the system.
=  S/Ninput / S/Noutput 

= 10 Log (Noise Factor)

FTotal = F1 + (F2 - 1) / G1 + (F3 - 1) /G1G2 + …

This shows that if the gain of the first stage is high, then the total noise figure will be approximately the noise figure of the first stage only. 

  • Sensitivity

Sensitivity = The input signal level required to produce a given S/Noutput

  • Minimum Detectable Signal

Minimum Detectable Signal = This is defined as the input voltage level corresponding to the sensitivity within a given measurement bandwidth.

In other words, this is the lowest level signal that can be detected above the noise floor.

  • 1 dB Compression Point

In connection with a mixer:

Normally, the output power is proportional to the input RF signal power.

However, if the input RF signal power becomes greater than the LO signal power, then the output power will become proportional to the local oscillator power.

Because the LO signal power is constant, the mixer output signal power will also be constant.

This point is called the saturation point.

The point where the output signal power is 1 dB down from where the ideal response would be is called the 1 dB compression point.

  • Single Tone Dynamic Range

1 dB Gain Compression Point – Minimum Detectable Signal

  • Two Tone Dynamic Range

(Two Tone) Third Order Intercept Point – Minimum Detectable Signal

  • Intermodulation Distortion (IMD)

Intermodulation Distortion (IMD) = The distortion created when two signals interact in a non-linear device and produce undesired higher order harmonics. The third order harmonics (which have the largest magnitude of all the harmonics) are (2f1 +/- f2) and (2f2 +/- f1).  Because the higher frequency signals (2f1 + f2) and (2f2 + f1) usually fall outside of the passband, they can be filtered out. Additionally, since the third order harmonic has a cubic exponent, the slope of the desired signal vs. the third order harmonic is 3 to 1 in dB. Thus, if the test signal increases by 1 dB, the IMD will increase by 3 dB, up until compression is reached.

 

wpe5.jpg (24599 bytes)

 

 

 

  • Two Tone Third Order Intercept Point (3OIP)

Two Tone Third Order Intercept Point (3OIP) = The point at which the Two Tone IMD signal intersects with a test signal.  As the 3OIP increases, so does the range of signals that do not create high IMD levels.  In a mixer, the 3OIP is usually referenced to the input signal as the test signal.  In an amplifier, the 3OIP is usually referenced to the output signal as the test signal.

The relationship between the Third Order Intercept Point, the power of the fundamental signals and the IMD level is as follows:

IP3 = Pout + (IMD (in dBc) /2)

where:
IP3 is the Third Order Intercept Point (in dBm).
Pout is the power of the fundamental signals (in dBm).
IMD is the amplitude of the Intermodulation Distortion (in dBc with respect to the amplitude of the incident fundamental signals).

 

  • Interference

As opposed to noise, which is present even in the absence of a signal, Interference can be viewed as any signal, or set of signals, other than the "desired" signal that manifests at the detector stage.

Interference can be classified many ways, including:

Adjacent Channel Interference

Co-Channel Interference (due to ISI or due to frequency reuse)

Intentional/Unintentional Jammers

etc...

 

In connection with Mixers:

Any undesired signal present at the "Image" frequency will manifest at the IF port and cause a spurious response.

The fundamental outputs of a mixer are two signals at sum and difference frequencies of the RF and LO.

V1(t) = A cos (wLO + wRF)t

V2(t) = A cos (|wLO - wRF|)t

Below is an example of "upmixing", where the LO is at a lower frequency than the RF signal.

(|wLO - wRF|) = wIF

 

Notice that spread symmetrically around the desired IF signal is a signal called the "Image" frequency that can also mix with the LO signal (in this case the sum component (wLO + wImage)) to produce a spurious signal at the IF.

Sometimes Bandpass filters centered around the RF signal can filter out the Image frequency. This is not possible if the Image frequency is very close to the desired RF signal, or if the Image frequency is in a band of frequencies of interest.  In this case, more complex (and expensive) Image Reject mixers can be used.

In addition, in a mixer, as the LO Drive Power is increased, the image rejection is improved (at the expense of LO-RF isolation).  Similarly, as the LO Drive Power is increased, the Intermodulation Distortion is decreased, thereby increasing the IP3 (again at the expense of Lo-RF isolation).

 

  • Impairment Effects on OFDM (Wireless LAN)  BER

As many Wireless LANs employ QAM over OFDM, it is illustrative to mention the effects of Power Amplifier non-linearities on the BER performance of OFDM.

Linear Power Amplifiers are required in the transceivers used for QAM over OFDM for several reasons.

Firstly, since QAM has a non-constant envelope characteristic, it necessitates the use of linear power amplifiers in the transceiver.   Secondly, the nature of the OFDM orthogonal signals tends to increase the peak to average ratio (dynamic range) of the signals fed into the amplifiers.  This also dictates the use of linear power amplifiers in the transceiver.

Any non-linearities in the transmit power amplifier will manifest as distortion (degraded BER for a given SNR) at the WLAN receiver.

One way to model the non-linear amplitude transfer characteristic of a power amplifier is by a cubic approximation as given in; "Impact of Front-End Non-Idealities on Bit Error Rate Performance of WLAN-OFDM Transceivers", by B. Come, et. al., Microwave Journal, February 2001, pp. 126.

 

 

  • Impairment Effects on Bluetooth communication Range

The range of any wireless communications link is dependent upon the path loss (a function of the operating frequency), transmit power, receiver sensitivity and Tx/Rx antenna gains.  Hostile impairments in the operating channel such as noise and interference also play a major role in determining the communication range.  Bluetooth is no exception.  The range of a Bluetooth device operating within a Bluetooth piconet is directly dependent upon the receiver sensitivity in the Bluetooth transceiver.  The receiver sensitivity is in turn dependent upon the phase noise of the I/Q demodulator local oscillator, or the phase noise of the PLL (if direct demodulation via a VCO is employed), etc. depending upon transceiver architecture.

The Bluetooth spec of -70 dBm Rx sensitivity is not very demanding, as cost and small size are primary drivers, and since the range is relatively short (<10 m).   However, as the trend moves toward integrating all Rx functions on one chip, digital noise (including the rise in phase noise due to employing PLLs to synthesis the LO on chip) can impact even a relatively relaxed spec like -70 dBm Rx sensitivity.

New and exotic semiconductor process technologies are being employed to accommodate low power, low noise, high bandwidth circuit topologies on a single chip to accommodate applications like Bluetooth.

 

The next installment in this series will explore smart antenna technology.