Power amplifiers (PAs) can be used in many different types of applications. For example, PAs are typically used in wireless devices such as cellular telephones and mobile Internet devices to amplify an outgoing radio frequency (RF) signal to a sufficient power level for transmission, e.g. to a cellular or other wireless network. Such power amplifiers often include one or more gain stages which can be formed of different types of semiconductor devices such as transistors.
In general, it is desirable for the PA to output a signal having a substantially constant gain over a temperature operating range of the device. However, certain PAs can suffer from large gain variations at different temperatures, if not supplied with a temperature dependent bias. As such, it is desirable to limit gain variation as the PA operates at different temperatures. One common method for compensation of gain as temperature changes is to use so-called PTAT currents, which are proportional to absolute temperature, to provide a bias current that increases with temperature. However, this temperature dependent current can negatively impact linearity characteristics of the PA. Thus a need exists for improved control of a PA or other amplifier to reduce gain variation.
According to one aspect of the present invention, a method includes generating a first current in a bias current circuit and biasing an amplifier with the first current when the amplifier is operating in a first temperature range, and generating a second current in the bias current circuit and biasing the amplifier with the second current when the amplifier is operating in a second temperature range. These two currents may correspond to different profiles with respect to temperature ranges. In some implementations, the currents can be generated with different weighted combinations of a constant current and a temperature dependent current.
Another aspect of the present invention is directed to a circuit for generating an output current that can be used as a bias current for an amplifier or other circuit. In one implementation, the circuit may include multiple differential pairs each including pairs of metal oxide semiconductor field effect transistors (MOSFETs). Each MOSFET of the pairs can be gated by a different temperature dependent voltage. Output terminals of one of the MOSFETs of each pair can be coupled to an output node to provide an output current according to a multi-level profile with respect to temperature. This profile may have different slopes in different temperature ranges. A voltage generator can generate the temperature dependent voltages, which may correspond to a differential voltage that varies with respect to temperature.
Yet another aspect is directed to a system having a transceiver to receive baseband signal information from a baseband processor and to output a radio frequency (RF) signal, an amplifier to amplify the RF signal, and a current generator to generate a bias current for the amplifier. As discussed above, the bias current can be temperature dependent and have multiple profiles in different temperature ranges. In this way, a reduced variation of a gain characteristic curve with respect to temperature can be realized, while maintaining substantial linearity.
In various embodiments, a technique may provide a temperature dependent bias for an amplifier, e.g., a gain stage of a PA. By providing a temperature dependent bias, the output of the gain stage may have reduced gain variation across a temperature profile at which the device may operate, while maintaining desired gain characteristic curves which are related to linearity. Specific circuits for providing this temperature dependent bias will be discussed further below.
However, a discussion of an example amplifier and the gain variation that is possible over a temperature profile first discussed. Referring now to
The quiescent current is defined as the current flowing when VIAC equals 0. In this case, the current value is determined by the NMOS and PMOS DC gate voltages. This quiescent current, along with device sizes and characteristics, sets the small signal gain of output stage 10. Because stage 10 is intended to operate as a large signal amplifier, as well as small signal, the devices will “self bias” as VIAC increases. That is, as input voltage amplitude increases, average current increases from the DC/quiescent value. The self bias is related to the nonlinear I/V characteristics of the MOSFET device.
Specifically, as signal (VIAC) amplitude increases, the nonlinear (super-linear) device characteristics cause the effective transconductance (GM) to increase. This effect is especially pronounced for devices with quiescent VON that is very small relative to input signal swing. This causes the output stage gain to increase as signal swing increases. During roll off after gain peaks, as the input amplitude further increases, output clipping occurs, where the stage begins to saturate in a classic amplifier sense, as devices begin to approach/enter the triode region, and gain reduces.
When taking all of these effects into account, a gain stage may be designed to have the gain characteristics (gain vs. input amplitude), shown in
With a basic understanding of these gain curves/characteristics, we can now begin to investigate the temperature effect on gain and linearity. For MOSFETs, in general the greatest effects of temperature are on mobility (μ) and threshold voltage (VT). Basic bias techniques can remove the effects of temperature on threshold voltage. This leaves the impact of mobility, which reduces at higher temperatures, resulting in gain reduction at higher temperatures. This gain reduction with respect to temperature can be difficult to resolve, as it would require temperature dependent calibration, which can be difficult to implement (unlike process variation, which can be removed with a one-time calibration).
Note that gain profiles with respect to input voltage (expansion, peaking, compression) such as seen in
As described above, a PTAT bias current can be used to compensate for temperature changes. Such a PTAT current may be generated on chip in a manner such that as the input temperature increases, so too does the PTAT current. This PTAT current can be used to counter the mobility reduction in the following gm equation:
gm=√{square root over (2IμCow/L)} [EQ. 1]
where I is a bias current designed to increase with increasing temperature in the case of a PTAT current, μ is the mobility, which decreases with temperature, C0 is the oxide capacitance, and W/L is a measure of the device size.
Although a PTAT bias technique can be used to control the small signal gain with temperature, this technique can have a significant impact on the gain curve characteristics. At large signal operation, the average Gm (related to gain) becomes determined more by drive conditions (VIAC) than by VON (i.e., VIAC>>VON). In this case, it makes less difference whether the quiescent currents are different between a PTAT current or a constant current case, as the Gm/gain values for these cases tend to converge.
When using a PTAT current, however, much larger peaking may occur, particularly at lower temperatures, as the device is basically starved of current at quiescent, and has a large signal gain that is similar to a constant current bias (due to the self bias). This behavior of a PTAT bias circuit can have a negative impact on linearity, as the excessive expansion results in significant distortion.
Thus in various embodiments a trade-off may be implemented between biasing a gain stage with PTAT and constant currents. This trade-off may maintain an acceptable level of linearity, while decreasing small signal gain variation. In this way, a bias profile can be implemented that does not diminish the quiescent current significantly at cold temperatures but still boosts current (to maintain gain) at hot temperatures. Thus in various embodiments, the trade-off may be made to favor linearity at the expense of slight gain variation. However, the gain variation using a multi-level current profile in accordance with an embodiment of the present invention may be relatively small, e.g., on the order of approximately 1.5 dB, while maintaining linearity in a range sufficient to pass requirements for applications such as a PA.
While the scope of the present invention is not limited in this regard, examples of a multi-bias current profile approach are shown in
In sum, the aggressiveness or slope of the current variation with temperature can be reduced at lower temperatures to preserve linear operation. Generally, this would be done by allowing the gain to be maintained as constant as possible as temperature goes lower (bias current would reduce) until the linearity limit is reached. As temperature goes higher, the current profile can be more aggressive (i.e., higher slope with temperature) as the VON will increase, and the operation will become more class A (with linearity not negatively impacted).
Different techniques may be provided to create this type of bias temperature profile by mixing a plurality of individual currents as a function of temperature. Each of these currents can have desired characteristics for different temperature ranges, e.g., cold and hot, respectively. For example, in one embodiment two individual currents may be mixed, with one current having characteristics (e.g., a lower or zero slope) for a first temperature range (e.g., a low temperature range) of between approximately −40 and 27 degrees Celsius (C) and the second current may have characteristics (e.g., a higher slope) for higher temperatures (e.g., higher than approximately 27° C.). While only two such currents are described for ease of illustration, embodiments may combine more than two currents each associated with a different temperature range. In one embodiment, the temperature may be determined using an on-chip sensor, which may be obtained from a bandgap circuit.
Referring now to
As further seen in
As further seen in
Referring now to
As seen in the embodiment of
As seen, differential pair 210 is biased with a current having a relatively low slope with respect to temperature (75% constant current, 25% PTAT current in the embodiment of
At very low temperatures (e.g., −20° C.), voltage generator 230 will cause VM to be larger than VP (the difference may vary depending on DC characteristics (Vo/Vi) and characteristics of the differential pair), and:
At high temperatures, voltage generator 230 causes VP>>VM, and:
In one embodiment, the circuit may be set such that at TΔ=0 (nominal temperature, e.g., T=27° C./300K), VP=VM (implies RM=RP), and the circuit is balanced (mismatch effects aside), as shown in
Although the bias/temperature profiles described above with specific reference for the hot and cold cases where a differential pair is saturated, in the transition region (around TΔ=0) the behavior is determined by the large signal characteristics of the devices, and the slope of VP−VM. Note that there need not be an instantaneous switch when temperature varies from an ambient, as in a transition region both differential pairs can be in a transition region, both providing current that is combined.
In an embodiment in which there are multiple gain stages, a single current generator may be provided, although the scope of the present invention is not limited in this regard. Note that in many implementations, the current profiles need not be fixed, and can be programmable, e.g., by varying the slopes. In addition, programmability can be realized by changing the temperature ranges (by the voltage generator output) for the operative profiles. In one embodiment, such control can be via control of the weighting coefficients (α and σ). In some embodiments, the programmability can be realized by off-chip control, e.g., from a baseband processor or a transceiver.
In different embodiments, a bias current generator can be coupled to a gain stage in different manners. In one embodiment, a magnetic coupling via transformers may be implemented. Referring now to
As described above, bias current generation circuitry in accordance with an embodiment of the present invention can be implemented in a PA such as a CMOS PA. Such a device can be used in various wireless systems, including handsets, mobile devices, PDAs and so forth. Referring now to
In a receive path, antenna 560 couples through antenna switch 555 and possibly through the duplexer or SAW filters and then to transceiver 540, which may demodulate the incoming RF signals back to baseband for transmission to baseband processor 530 for further processing. While shown with this particular implementation in the embodiment of
While the present invention has been described with respect to a limited number of embodiments, those skilled in the art will appreciate numerous modifications and variations therefrom. It is intended that the appended claims cover all such modifications and variations as fall within the true spirit and scope of this present invention.
Number | Name | Date | Kind |
---|---|---|---|
4317184 | Leszczynski | Feb 1982 | A |
4633165 | Pietkiewicz et al. | Dec 1986 | A |
4789819 | Nelson | Dec 1988 | A |
5608347 | Kearney | Mar 1997 | A |
6507238 | Yang | Jan 2003 | B1 |
6512412 | Casper | Jan 2003 | B2 |
6566849 | Chavan et al. | May 2003 | B1 |
6940339 | Otaka | Sep 2005 | B2 |
7991437 | Camuffo et al. | Aug 2011 | B2 |
20040169537 | Abbasi et al. | Sep 2004 | A1 |
20090243727 | Bockelman et al. | Oct 2009 | A1 |
20100093291 | Embabi et al. | Apr 2010 | A1 |
Entry |
---|
U.S. Appl. No. 12/317,819, filed Dec. 30, 2008, entitled “An Output Gain Stage for a Power Amplifier,” by Vishnu Srinivasan, et al. |
Number | Date | Country | |
---|---|---|---|
20110086594 A1 | Apr 2011 | US |