GB2376822A

GB2376822A - A power amplifier using a measure of current drawn to produce an error signal which in turn is modified to produce a gain control signal

Info

Publication number: GB2376822A
Application number: GB0115401A
Authority: GB
Inventors: Jonas Persson
Original assignee: Telefonaktiebolaget LM Ericsson AB
Current assignee: Telefonaktiebolaget LM Ericsson AB
Priority date: 2001-06-22
Filing date: 2001-06-22
Publication date: 2002-12-24
Anticipated expiration: 2021-06-22
Also published as: GB2376822B; GB0115401D0

Abstract

A power amplifier gain control loop comprising a measurement unit, an error detection unit and a signal processing unit. A measurement x<SB>5</SB>(t) indicative of the amount of current drawn by the power amplifier 1 is made and used, preferably with a reference signal x<SB>1</SB>(t)which may be retrieved from a memory, to produce an error (or difference) signal x<SB>2</SB>(t) which is modified in a predetermined way, in the signal processing unit, to produce a gain control signal. The predetermined modification of the error signal may include filtering and/or predistortion. The problem addressed is that of adjusting telephone parameters so that they fit the power vs. time template required by, e.g., GSM or any other system requiring accurate power control (TDMA or CDMA systems). This in turn (page 4) leads to easier implementation of the required power vs. time mask.

Description

-1- 2376822

TELECO I=TIONS SYSTEMS

The present invention relates to telecommunications systems, and in particular to power 5 control in mobile telephone systems.

BACKGROUND OF THE INVENT ION

The requirements for output power control in 10 mobile telephones can often be difficult to achieve.

The requirements for GSM, for example, can be found in the ETSI specification " 05. 05 Digital cellular

telecolfununi cations system; Radio transmission arid reception". There are three critical parameters 15 concerning the transmitter output power.

À Output power level during a constant power part ("mid part") of a transmitted burst.

Power vs. time, i.e. output power during up 20 ramping and down-ramping parts of a transmitted burst. À Spectrum due to switching (up- and downramping).

Several output power classes are specified in the 25 05.05 document. These power levels should be kept within well-defined tolerances.

The "power vs. time" requirements state that the transmitted power should fit within a specified template, of output power versus time. The template 30 can be illustrated as a graph of power vs. time.

Adjusting the telephone parameters so that they fit the power vs. time template can be a very time-consuming task during development and critical during manufacturing. 35 The spectrum due to switching requirement means

-2- that the spectrum caused by the ramping (switching) process should fit in a specified spectrum mask. It is therefore necessary to have a "good" (reliable) power vs. time behaviour, not only to fulfil the power vs. 5 time template but also to avoid spectrum contamination.

It is to be noted that, although the GSM system is used as an example, the ideas presented in this specification could be used in any TDMA (Time Division

Multiple Access) system, or any system that requires 10 fast and/or accurate power control, such as CDMA.

Arranging the power control so that the telephone fits the power vs. time template and the spectrum due to switching mask, can be a very timeconsuming task.

In production, good yield is necessary.

15 In Figure 1 of the accompanying drawings, the principle of today's power control solution is shown.

A power amplifier 1 is connected to receive an RF input RF:n The power amplifier operates to output an amplified RF signal RFout to an antenna 2, as is known 20 and understood.

In order to control the power output of the power amplifier, the current, Ic, used by the PA (Power Amplifier) 1, is measured (through a resistor R) . This current value provides an indirect measurement of the 25 PA output power. The measurement provided by voltage x4=RIc, is fed back for comparison with an input control signal x1. A difference (or error signal) , x2, is filtered by a loop filter, Hip, to produce a control signal X3, which is used for controlling the PA RF 30 output power. Signal X3 is often called Vapc (ape-amplifier power control). Ideally, the measurement signal X4 should track the input control signal x.

The total transfer function for the control system 35 (Htot=x4/x) can be found from the following:

-3- X4 = X3HPA = X2HI,PHpA = (X! - x4)HL,pHpA (l) 5 X4(1+ HLP0PA) = XIHL. PHPA (2)

Hto,_ X4 = HLPHPA X 1+ HtcpHpA Minimizing the difference between x1 and x would provide an ideal control loop. This means that X4 /X1 15 or H pHpA> 1.

Ideally, the transfer function HPA =X4/X3 should be constant(= IC/Vapc). However, in practice, this is not generally the case. As illustrated in Figure 2, the transfer function of the feedback loop typically 20 varies, i.e. the feedback loop gain varies. This variation is due to the variation of the PA transfer function HPA with the control voltage Vapc. Thus, the maximum achievable error reduction of the control system will vary. In the Figure 2 example, the loop is 25 practically "open" for low VapC and high VapC values, causing poor tracking ability in the control system.

For medium Vapc values however, the tracking ability is good, since the loop gain is high.

The non-constant behaviour of HPA will also result 30 in implementation difficulties for the loop filter since the risk of instability is high. The reason for this is that the loop filter must have sufficient gain to achieve good error reduction and fast control even at low or high VapC values (where HPA is small). On the 35 other hand, this means increased risk for instability

-4- at medium VapC values (where HPA is large).

SUMMARY OF THE PRESENT INVENTION

5 The invention presented in this document adds a biasing pre-distortion block to the control loop shown in Figure 1. By doing this, the behaviour of the PA control loop will have less loop gain variation, since the gain variations of HPA is compensated for.

10 Distinguishing properties of the presented solutions are: The power vs. time mask (in GSM specified in 05.05) should be more straightforward to fulfil, since variations in the loop gain due to variations in HPA are 15 reduced. This will make implementation of suitable loop filters easier.

Since power vs. time will be easier to control, this will also mean that it is easier to do the up ramping and down-ramping in such a way that the 20 spectral contamination is held low. Thus, the requirements on spectrum due to switching (in GSM specified in 05.05) will be easier to fulfill.

Although the GSM system is used as an example, the ideas presented in this report could be used in any 25 TDMA (Time Division Multiple Access) system, or any systems that require fast and/or accurate power control. It is emphasized that the term "comprises" or "comprising" is used in this specification to specify

30 the presence of stated features, integers, steps or components, but does not preclude the addition of one or more further features, integers, steps or components, or groups thereof.

35 BRIEF DESCRIPTION OF THE DRAWINGS

-5- Figure 1 illustrates a prior art power amplifier

and control circuit; Figure 2 illustrates a transfer function of the control circuit of Figure 1; 5 Figure 3 illustrates a power amplifier and control circuit embodying one aspect of the present invention; Figure 4 illustrates respective transfer functions of parts of the circuit in Figure 3; Figure 5 illustrates a derivative of one of the 10 transfer functions shown in Figure 4; Figure 6 illustrates control loop gain in a i practical example of the circuit of Figure 3; Figure 7 illustrates a power amplifier and control circuit embodying another aspect of the present 15 invention; and Figure 8 illustrates a power amplifier and control circuit embodying another aspect of the present invention. 20 DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Figure 3 is a block diagram of a power amplifier and control circuit embodying one aspect of the invention. The Figure 3 embodiment is similar to the 25 circuit of Figure 1. However, the Figure 3 circuit includes an extra functional block, Hpre, in the feedback control loop. The extra block Hpre introduces an additional term in the transfer function of the feedback control loop. The overall transfer function 30 of the feedback control loop can be found from the following, with reference to Figure 3:

-6- X5 = X4HPA

= X3HL,pHPA = X2HpreHLPHRA 5 = (x! - X5)HpreHLPHP t (4) Xs (1 + HpreHLPHPA) = x; HpreHLPHPA (5) 10 Which gives: H = X5 = HpreH PHPA (6) To produce an ideal feedback control loop, the difference between x1 and Xs should be minimized. This means that x5/x1=1 or in other words HpreH pHpA>>1.

As discussed above, HPA is not constant. However, 20 it is desirable to make the transfer function Htot=x5/x constant. If the loop filter (HAP) is assumed to be linear (i.e. the gain is independent of input signal), Htot can be made constant by choosing Hpre=kHpA-1, where k is a constant. With Hpre= kHpa 1in equation (6), this 25 gives: Htot- kHLp (7) In other words, by introducing a pre-distortion block, Hpre=kHp^1, in the feedback control loop, the overall loop gain can be made to be constant (i.e. independent of the power amplifier Vapc=x4). Figure 4 35 illustrates such ideal pre-distortion using Hpre. In

-7- the ideal case, Hpre x4 /X5. This can be seen to mean that Hpre is proportional to the inverse of the derivative of the function Ic vs. Vapc The function and its derivative are illustrated in Figure 5.

5 However, it is not always necessary to eliminate completely variations in the loop gain by achieving perfect pre-distortion. In a solution for mobile telephones, for example, it could be acceptable to use an implementation that simply reduces the loop gain 10 variations by a desired amount. Figure 6 illustrates the PA transfer function (HPA) and the resulting overall transfer function (HpreHpA) following use of the pre distortion block in a practical embodiment.

The pre-distortion function Hpre can be implemented 15 or calculated in several ways.

For example, by varying the gain HPA Of the PA 1 in real time, a circuit solution in the analogue domain can be used to determine Hpre. An analogue circuit which has X4 and X5 as inputs can be used to determine 20 Hpre such that variations in HTOT are reduced. The signal "gain control" in Figure 7 sets the gain of Hpre.

Alternatively, Hpre can be calculated in the digital domain whilst varying the HPA in real time.

Again, both X4 and X5 are used as inputs for deciding 25 proper gain control signal to supply to the pre distortion block. The gain control signal is calculated to minimise variations in HTOT Alternatively, a burst-based learning solution can be used. In such a case, the gain of the predistortion 30 block is first set constant (=1) during one burst.

Values of X4 and X5 are sampled (collected) during this burst. The PA characteristic, HPA, is thereby obtained and a suitable Hpre{x5} (or Hpre{x4}) is then used during the call, for all bursts that have the same nominal 35 ("mid-part") power as the one that the Hpre{x5} (or

-8- Hpre{x4}) was meant for. When a "new" power level is requested for the first time during a call, the procedure is repeated, ie. X4 and xs are collected and a Hpre{xs} (or Hpre{x4}) for this power level is calculated.

5 Hpre{x5} (or Hpre{x<}) for different power levels are stored in a memory (table). When the Hpre{x5} (or Hpre{x4}) is to be used, first the memory address containing the table is addressed to find the data Hpre{Xs} (or Hpre{x4}) associated with the power level lo that is to be sent. Then, during transmission of the burst, the elements in the table (on the memory address) are addressed with xs (or X4).

As an alternative x1 could be used to address the elements in a table containing Hpre{x1} The learning 15 principle is the same as above, with the exception that not only values of X4 and xs are sampled (collected) during the burst but also x1.

As a further alternative, a learning procedure can be used during manufacture of the device concerned 20 (telephone, for example). Tables, addressed with X4 and/or X5, are used for deciding proper gain in the prep distortion block. For each power level (and with Hpre=constant=1), values of X4 and X5 are sampled (collected). Thus, the PA characteristic, HPA, is 25 obtained and a suitable Hpre{x5} (or Hpre{X4}) for each power level can be computed. Hpre{x5} (or Hpre{X4}) for different power levels are stored in a memory (table).

When the Hpre{x5} (or Hpre{x4) is to be used, first the memory address containing the table is addressed to 30 find the data Hpre{x5} (or Hpre{x} ) associated with the power level that is to be sent. Then, during transmission of the burst, the elements in the table (on this memory address) are address with xs (or X4).

As an alternative, x1 could be used to address the 35 elements in a table containing Hpre{x1}. The "learn-up"

- 9 - principle is the same as described above, with the exception that not only values of xs and xs are sampled (collected) during the burst, but also x1.

The learning procedure can also be used to produce 5 tables for reference during use. The gain Hpre in the pre-distortion block then depends on what power level the telephone is requested to transmit on during the "mid part" of the burst.

Figure 7 shows the principle of using the real 10 time or burst variation of the Hire, to reduce loop gain variations due to gain variations in HPA.

In Figure 8, the principles of using the tabular methods are illustrated. The gain of Hpre is varied in a manner that has been determined by the learning 15 procedure described-above.

Reducing variations in feedback loop gain means that the power vs. time mask in TDMA systems (in GSM this is specified in the 05.05 ETSI document) will probably be easier to fulfil. This will make 20 implementation of suitable loop (filter) easier.

Since power vs. time will be easier to control, up-ramping and downramping will be possible in such a way that the spectral contamination remains low. Thus, the requirements on spectrum due to switching (in GSM 25 this is specified in the 05.05 ETSI document) will probably be easier to fulfil.

Claims

-10 CLAIMS:

1. A power amplifier circuit comprising: a power amplifier having a control input, a signal input and a signal output, the amplifier having a gain 5 value determined by a control signal applied to the control input of the amplifier; a measurement unit operable to produce a measurement signal indicative of an amount of electrical current drawn by the power amplifier; and 10 a control unit comprising: an error detection unit operable to receive the measurement signal from the measurement unit, and operable to produce an error signal; and a signal processing unit operable to receive the 15 error signal from the error detection unit, and to modify the error signal in a predetermined manner to produce a gain control signal for supply to the power amplifier as the said control signal.

2. A circuit as claimed in claim 1, wherein the 20 predetermined manner of modification of the error signal is determined as a function of the error signal and the measurement signal.

3. A circuit as claimed in claim 1 or 2, wherein the error detection unit is operable to receive a 25 reference signal and is operable to produce the error signal in dependence upon the reference signal and the measurement signal.

4. A circuit as claimed in claim 3, wherein the reference signal is determined by stored reference 30 values.

5. A circuit as claimed in claim 3 or 4, wherein the predetermined manner of modification of the error signal is determined by a function of the error signal, the measurement signal and the reference signal.

35

6. A circuit as claimed in any one of the

preceding claims, wherein the signal processing unit is operable to distort the error signal so as to produce a gain control signal which substantially corrects for gain variations occurring in the power amplifier 5 circuit.

7. A circuit as claimed in any one of the preceding claims, wherein the predetermined manner of modification is selected by determining the error signal during a first burst of output power from the 10 power amplifier.

8. A circuit as claimed in any one of the preceding claims, wherein the predetermined manner of modification of the error signals is determined by reference to stored values.

15

9. A circuit as claimed in claim 8, wherein the signal processing unit is operable to access stored values on the basis of a combination of the reference signal, the measurement signal and the error signal.

10. An RF power amplifier circuit comprising: 20 a power amplifier having RF signal input and output terminals and a gain control signal input, the amplifier having a RF gain value determined by a gain control signal received at the gain control signal input; and 25 a control circuit for providing a gain control signal to the power amplifier, wherein the control circuit comprises: current measurement means operable to provide a measurement signal indicative of an amount of 30 electrical current drawn by the power amplifier, error detection means operable to produce an error signal in dependence upon the measurement signal and a reference signal; and processing means operable to receive the error 35 signal from error detection means and to apply a

-12 predetermined distortion to the gain control signal, the power amplifier, current measurement means, error detection means and processing means forming a feedback loop having a transfer function, the 5 predetermined distortion applied by the processing means to the error signal serving to reduce variations in the transfer function of the feedback loop.

11. A circuit as claimed in claim 10, wherein the predetermined distortion of the error signal is 10 determined as a function of the error signal and the measurement signal.

12. A circuit as claimed in claim 10 or 11, wherein the error detection means is operable to receive a reference signal and is operable to produce 15 the error signal in dependence upon the reference signal and the measurement signal.

13. A circuit as claimed in claim 12, wherein the reference signal is determined by stored reference values. 20

14. A circuit as claimed in claim 12 or 13, wherein the predetermined distortion of the error signal is determined by a function of the error signal, the measurement signal and the reference signal.

15. A circuit as claimed in any one of claims 10 25 to 14, wherein the processing means is operable to distort the error signal so as to produce a gain control signal which substantially corrects for gain variations occurring in the power amplifier circuit.

16. A circuit as claimed in any one of claims 10 30 to 15, wherein the predetermined distortion is selected by determining the error signal during a first burst of output power from the power amplifier.

17. A circuit as claimed in any one of claims 10 to 16, wherein the predetermined distortion of the 35 error signal is determined by reference to stored

-13 values.

18. A circuit as claimed in claim 17, wherein the processing means is operable to access stored values on the basis of a combination of the reference signal, 5 the measurement signal and the error signal.

19. A method of controlling the output power of a power amplifier circuit, the method comprising: measuring an electrical current drawn by a power amplifier; 10 producing an error signal in dependence upon the measured current and a reference signal; applying a predetermined modification to the error signal to produce a gain control signal which reduces variations in the transfer function of the feedback 15 loop.

20. A method as claimed in claim 19, wherein the predetermined modification is determined as a function of the measured current and the gain control signal.

21. A method as claimed in claim 19 or 20, 20 wherein the predetermined modification is determined from a look up table containing known modification factors.

22. A method as claimed in claim 19, 20 or 21, wherein the reference signal is determined by stored 25 reference values.

23. A method as claimed in claim 19, 20, 21, or 22, wherein the predetermined modification of the error signal is determined by a function of the error signal, the measurement signal and the reference signal.

30

24. A method as claimed in any one of claims 19 to 23, wherein the predetermined modification is selected by determining the error signal during a first burst of output power from the power amplifier.

25. A method as claimed in any one of claims 19 35 to 24, wherein the predetermined modification of the

-14 error signal is determined by reference to stored values.

26. A method as claimed in claim 25, conmprising accessing stored values on the basis of a combination 5 of the reference signal, the measurement signal and the error signal.