Amplifiers – With semiconductor amplifying device – Including gain control means
Reexamination Certificate
2000-11-14
2004-08-31
Pascal, Robert (Department: 2817)
Amplifiers
With semiconductor amplifying device
Including gain control means
C330S252000, C330S254000
Reexamination Certificate
active
06784741
ABSTRACT:
The invention relates to a low noise amplifier having variable gain.
Low noise amplifiers in cellular radios are optimised for a variety of performance and cost trade-offs. Conventionally, these are fixed gain circuits, and the AGC function of the radio is effected in the IF stages. As a result the intermodulation performance of the radio is mainly set by the front end performance, and with a strong signal at the input, the IF stages may well be operating at much reduced gain, but the LNA and mixer may be driven beyond their limits of linearity.
To improve the overall dynamic range of a receiver it would be useful to balance the gain and linearity differently at the input under strong signal conditions. One possibility is to reduce the gain of the LNA. If the input referred third order intermodulation product (IP3) level also goes up when the gain is reduced, then the performance limiting stage is likely to be the mixer. However, if the input referred IP3 level of the LNA is not improved, then the LNA will still generate distortion products before the mixer, and no real improvement in dynamic range will be achieved.
A typical design for a BICMOS LNA with a differential signal path is a simple cascoded common emitter stage. The load may be resistive or inductive or a combination of the two. An inductance in the emitter circuit is advantageous as it raises the real part of the input impedance without increasing noise. This is important as the amplifier in combination with the package lead inductance, must provide a good termination for the off-chip band-select filter. If the match is poor (i.e., with a return loss of more than about −12 dB) the filter function will be affected to an unacceptable degree. If resistive degeneration is used in the emitter circuit, the input capacitance increases as well as the real part, and there is more noise.
In a discrete component LNA gain can be controlled simply by the inclusion of a passive attenuator in the input path. A simple T or &pgr; attenuator with PIN switching can produce the effective gain change and keep the match as good or better in the reduced gain state. The signal loss obviously affects the noise figure in the low gain setting but this is not such a big problem. To make this work in a silicon monolithic form is more difficult. MOS switches are apparently ideal for selecting such an attenuation function, but in practice there are big problems. To get a sufficiently low insertion loss the channel resistance must be low compared with 50 &OHgr; (or 100 &OHgr; differential), and so channel widths must be large. Consequently, there will be large capacitances to substrate leading to loss in either the ON or OFF states. Hence the noise figure is poor in the high gain setting.
A known way of controlling the gain in an integrated LNA is to use a current steering architecture as disclosed in a paper entitled “Dual Bond High-Linearity Variable-Gain Low-Noise Amplifier for Wireless Applications,” by Keng Leong Fong and published in the Proceedings of 1999 IEEE International Solid State Circuits Conference at pages 224 and 225. The input transistor is split into sections having a ratio related to the attenuation step required. The cascode transistors are also split in the same ratio. One of the paths, however, is provided with a diversion path activated by another cascode operated as a steering switch. The bias on this steering point is switched from about 400 mV below the cascode bias (high gain setting) to about 400 mV above the cascode bias (low gain setting). Hence in the low gain setting only a proportion of the signal current reaches the load. The signal conditions, viewed from the input port, remain the same and so the match to the input is unaltered. At first sign there is no deterioration in noise figure, but in reality the split input transistors act as a differential pair without degeneration in the low gain state, and so some additional noise reaches the output without being cancelled.
This arrangement offers a variable gain without causing the input impedance to change but because the input transistor operating points are essentially unchanged, there is no improvement in the linearity performance. If the mixer input referred third order intermodulation product point is higher than that of the LNA it is still worth having a modest gain step implemented in this way, but for more dynamic range some improvement is desirable.
A possible way of increasing the range of gain control is to use resistive emitter degeneration which can be switched into and out of operation. This, however, has the problem that it alters the input impedance of the amplifier and consequently adversely affects the matching between the RF filter and the amplifier input.
It is an object of the invention to enable the provision of a low noise amplifier in which one or more of the disadvantages discussed above are mitigated.
The invention provides a low noise amplifier having switchable gain settings, the amplifier comprising a cascoded common emitter amplifier stage, a first arrangement for reducing the gain to an intermediate level, said arrangement including a signal diversion path which when activated, causes a portion of the output signal to be diverted from the output and dumped, a second arrangement for reducing the gain to a low value, said second arrangement comprising means for increasing emitter degeneration and means for connecting a shunt feedback path from the diversion path to the input of the amplifier to reduce the increase in input impedance caused by the increase in emitter degeneration.
The common emitter and the cascode may both comprise multiple transistors, said first arrangement comprising means for diverting the current in one of the transistors making up the multiple transistors into the diversion path.
The first arrangement may comprise a further transistor connected to the junction of the common emitter and cascode transistors, said further transistor being operable in an intermediate gain state of the amplifier to conduct the signal current of the one transistor of the common emitter stage to a supply rail.
The second arrangement may comprise a resistor in the emitter circuit of the common emitter transistor and a switch connected in parallel with the resistor, the switch being opened to connect the resistor in the emitter circuit.
The second arrangement may further comprise a still further transistor connected to the junction of the common emitter and cascode transistors, said still further transistor being arranged in the low gain state of the amplifier to apply the output current of the one common emitter transistor to an internal load, a feedback path being provided between the internal load and the input of the amplifier.
An amplifier according to the invention has several advantages. First, because the feedback network is connected from an amplified output, the resistors used can have a resistance several times larger than those needed in a termination connected to ground and hence their intrinsic noise contribution is lower. Secondly, if the network contains larger value resistors the MOS switches can be physically smaller, since their ON resistance needs only to be low in comparison with the resistors and not the absolute input impedance. Consequently, their parasitic capacitance to ground is lower making the path for noise in the OFF state less significant. Thirdly, by using the dummy output to drive the feedback path there is no other loading at this point and the gain and phase are not affected by the load presented by the input capacitance of the following circuit (as is the main output). A further advantage of this approach is that the variable gain and improved linearity are obtained with no significant additional power consumption. Only some bias for the current steering path needed.
The low noise amplifier may comprise a differential pair. This enables the production of a differential amplifier using the same inventive principle. This will, of course, mean that the various diversion means and feedback paths are
Nguyen Patricia T.
Pascal Robert
Waxler Aaron
LandOfFree
Amplifier does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Amplifier, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Amplifier will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3313252