Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.

Broadband active mixers and in-phase/quadrature (I/Q) demodulators are available with excellent performance across more than three octaves. This equips them for receiver designs aimed at broadband applications and multiple frequency bands. Optimum performance with double-balanced mixers and I/Q demodulators can be achieved by driving them with differential, rather than single-ended, signals. This necessitates the use of baluns on the mixer RF and local-oscillator (LO) ports.

Single-ended drive without a balun is an option, but results in degraded performance. Narrowband baluns can provide low insertion loss and good return loss, but with limited frequency coverage. Operating an active mixer with a low-loss broadband balun on the local-oscillator (LO) and RF ports is the most desirable option when designing a broadband or multifrequency receiver platform. What follows will compare the performance of an active RF mixer and I/Q demodulator when driven with differential signals and with single-ended signals.

Broadband wireless receiver designs tend to fall into one of two categories: wideband or frequency-selective types. Wideband receivers can be tuned over a frequency range of several octaves, with relatively frequency-agile tuning. Reconfigurable radio designs, on the other hand, are hardware platforms that can be easily adapted to work at different frequencies. For example, a receiver that is deployed in two different geographical regions would have the same core receiver design. However, each would be built and configured to accommodate a particular wireless standard.

The analog component building blocks with these receivers that vary from system to system include filters, low-noise amplifiers (LNAs), and voltage-controlled oscillators (VCOs). The core receiver circuitry employs a common design for multiple frequency bands and standards. Development of reconfigurable radio designs is cost effective because the end result is a single radio platform which can be reused in different frequency deployments. Additionally, from a manufacturing perspective, this reduces the diversity components that must be stocked.

Models ADL5801 and ADL5802 active mixers and model ADL5380 I/Q demodulator from Analog Devices were specifically designed to serve the needs of these systems. The ADL5801 and ADL5802 are single- and dual double-balanced active mixers, respectively, designed to operate from 10 MHz to 6 GHz. They exhibit single-sideband (SSB) noise figure of 9.75 dB and an input third-order-intercept point (IP3) of +28.5 dBm at 1900 MHz. The active mixers operate on a single supply of +5 VDC and have adjustable bias for low power operation. Model ADL5380 is an I/Q demodulator which operates from 400 MHz to 6 GHz. It has an input IP3 of +28 dBm and a noise figure of 12 dB at 1900 MHz.

Ideally, the LO and RF ports of these devices should be driven differentially. This has implications for the support circuitry that is required to convert signals between single-ended and differential formats. Before considering this more closely, it may help to examine the operation of the double-balanced active mixer core used in these devices.1

Differential Drive Optimizes Active Mixers, Fig. 1

Figure 1 shows a transistor-level schematic diagram of a Gilbert-cell double-balanced active mixer. The term “double balanced” refers to the mixer’s LO and RF ports being driven differentially. The basics of a Gilbert cell mixer can be best described as a multiplication in the time domain of the input RF signal by a square wave, with value +1 or -1, at the LO frequency. Referring to Fig. 1, first consider what happens when the voltage at LO+ is positive and the voltage at LO- is negative, so that transistors Q1 and Q4 are turned on, while transistors Q2 and Q3 are turned off. This results in Q1 and Q4 behaving as closed switches with the outputs are taken at the IF ports, with the outputs multiplied by +1.

For the opposite scenario, where the voltage at LO- is positive and LO+ is negative, the IF outputs are now interchanged with respect to the previous case. In other words, the outputs have been multiplied by -1. This square wave with amplitude of +1 or -1 at the LO frequency is the mixing signal. It is the multiplication of the mixing signal with the RF signal which results in the sum and difference terms at the IF outputs. It may not be obvious, but the sum and difference terms reveal themselves when the RF and LO signals are expressed mathematically and the multiplication expanded.

Mathematically, the RF input voltage can be defined as VRF(t)cos(ωRFt) and the mixing signal is a periodic square wave which can be expanded as follows:

Vmix(t) = (4/π){cos(ωLOt) - [cos(3ωLOt)/3] + [5cos(ωLOt)/5] -…}

Multiplication of the RF and mixing signals yields:

V(t) = VRF × Vmix

V(t) = (2VRF/π)[cos(ωRFt – ωLOt)[cos(ωRFt + ωLOt)] - (2VRF/3π)[cos(ωRFt – 3ωLOt) +[cos(ωRFt + 3ωLOt)] +…

After applying filtering to reject higher-order harmonics, the resulting output of the mixer can be written as:

V(t) = (2VRF/π)[cos(ωRFt – ωLOt) + [cos(ωRFt + ωLOt)]

Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.