When interfaced with the AD8366, which has a 217-? differential input impedance, the ADL5380 has 5.9-dB voltage gain and –0.5-dB power gain [5.9 dB – 10log (217/50)]. For best performance, the common-mode voltage between the ADL5380 and AD8366 is set to 2.5 V by connecting the ADL5380 ADJ pin to VS. A differential fourth-order Butterworth low-pass filter with 0.5-dB insertion loss, placed between the ADL5380 and the AD8366, suppresses noise and unwanted high-frequency components. While the filter will cause some mismatch, it will be tolerable at baseband frequencies.
The common-mode output voltage of the AD8366 can be set to 2.5 V; it has best linearity when VCM is left floating. Unfortunately, the AD6642 has best performance with 0.9-V common-mode input voltage (0.5 × AVDD). Because the common-mode output voltage of the AD8366 must be between 1.6 V and 3 V, the AD6642 VCM and AD8366 VCM terminals cannot be connected directly, and resistors must be used to divide the AD8366 common-mode output voltage down to 0.9 V.
For best performance, the AD8366 should drive a 200-? load. To achieve the desired common-mode level and impedance match, 63-? series resistors and 39-? shunt resistors are added after the AD8366. This resistor network will attenuate power gain by 4 dB.
The AD8366 output can swing 6 V p-p, but the 4-dB attenuation provided by the resistor network limits the voltage seen by the AD6642 to 2.3 V p-p, protecting it from damage caused by big interference spikes or uncontrolled gains.
A differential sixth-order Butterworth low-pass filter with 1.5-dB insertion loss, placed between the AD8366 and the AD6642, filters unwanted high-frequency components. The complete differential interface for the I channel is shown in Figure 2.
Figure 2. ZIF receiver interface diagram and simulated filter characteristics.
To preserve enough margin to account for gain variation over temperature, the AD8366 gain is set to 16 dB for the normal mode.
In this configuration, the gain of the whole signal chain is
5.9 dB – 10log (217/50) – 0.5 dB + 16 dB – 10log (200/217) – 1.5 dB – 4 dB = 9.9 dB.
The two LNAs inserted in cascade ahead of the ADL5380 achieve 32 dB of gain. With the analog-to-digital converter configured for a 2-V p-p swing and 78-? equivalent input impedance, it is able to handle a –34-dBm single-tone RF input signal. If the input signal has a 10-dB peak-to-average ratio (PAR) when modulated, a –41-dBm input signal is the maximum signal that the receiver can handle without changing the VGA setting.
In the other words, voltage gain can be used to calculate the signal chain link budget. When the input port impedance is equal to that of the output port, the voltage gain is equal to power gain. The voltage gain of the whole signal chain is
32 dB + 5.9 dB – 0.5 dB + 16 dB – 1.5 dB – 8 dB = 43.9 dB.
For single-tone signal input, to get a 2-V p-p swing range, the proper input power is
8 dBm – 43.9 dB + 10log (78/50) = –34 dBm.
The result is a close match to the calculated power gain.
In some applications, the ADL5380 may need to be connected directly to the AD6642, in which case a 500-? resistor can be added to the AD6642 differential inputs to improve matching. The ADL5380 voltage gain will be 6.9 dB, with the same common-mode problem as with the AD8366. A 160-? series resistor and 100-? shunt should be used to achieve a 500-? load and the desired common-mode voltage. Again, the resistor network attenuates the voltage by 8 dB (and the power by 4 dB).
A low-pass filter with 1.5-dB insertion loss, placed between the ADL5380 and AD6642, filters unwanted frequency components. The input impedance is 50 ?, and the output impedance is 500 ?. In this configuration, the gain of the whole signal chain is
6.9 dB – 10log (500/50) – 1.5 dB – 4 dB = –8.6 dB.
Superheterodyne Receiver Interface Design and Gain Calculation ?In superheterodyne receivers, the system uses ac coupling, so the dc common-mode voltage does not have to be considered when interfacing these circuits. Many mixers, such as the ADL535x and ADL580x, have 200-? differential output impedance, so the power gain and voltage gain are presented separately for different output impedances.
Figure 3 shows one channel of a superheterodyne receiver implemented with an ADL5523 low-noise amplifier; an ADL5356 dual balanced mixer with LO buffer, IF amplifier, and RF balun; a low-pass filter; an AD8376 dual ultralow distortion IF VGA; another low-pass filter; and an AD6642 dual IF receiver.
Figure 3. Superheterodyne receiver diagram; one channel shown.
This design uses a 140-MHz IF and 20-MHz bandwidth, so the parts can be ac-coupled.
The AD5356 has best performance with a 200-? load, but the AD8376 has 150-? input impedance. Thus, to suppress mixer output spurs and provide better impedance matching, the differential LC filter must have 200-? input impedance and 150-? output impedance. In applications where the output band signal must be suppressed by a sharp filter, a differential SAW filter can be used, but this introduces loss and group delay in the receiver signal chain. A differential fourth-order band-pass Butterworth filter may be suitable for many wireless receivers because the RF filter can provide enough attenuation for out-of-band interference.
The AD8376’s current-output circuit has high output impedance, so 150-? is needed between its differential outputs. Another differential filter must attenuate the second- and third-harmonic distortion components, so this 150-? load is divided into two parts. First, a 300-? resistor is installed in the output of the AD8376. Another 300-? resistor is formed by two 165-? resistors and the ADC’s 3-k? input impedance. The two 165-? resistors also provide the dc common-mode voltage for the ADC input. The LC filter’s input and output impedances are both 300 ?. Perfect source and load matching is very important for high-IF applications. The complete interface is shown in Figure 4.
Figure 4. Superheterodyne receiver interface diagram and filter simulation result.
In the receiver, a 20-dB LNA is installed ahead of the mixer. The filter after the mixer has 2-dB insertion loss; the filter between the AD8376 and the ADC has 1.2-dB insertion loss. The AD8376 gain is set to 14 dB to provide enough margin to account for temperature variation. The overall gain of the receiver is
20 dB + 8.2 dB – 2 dB + 14 dB – 1.2 dB = 39 dB.
To limit the ADC input voltage to less than 2 V p-p, the power transmitted to the 150-? resistance (300 ? || (165 ? × 2) || 3 k ?) should be smaller than 5.2 dBm. The maximum input power for the receiver is thus –33.8 dBm for a single-tone signal. If the input signal is a 10-dB PAR modulation signal, the maximum input signal using this gain setting is –40.8 dBm.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.