Calorimetric RF power meter with dynamic zeroing and constant
Contents
1. A A D 1 72 m 53 BUFFER SLIGHT RUN COMPARISON 55 AMP i OFFSET POWER 9 NN ADC MEASUREMENT 61 RESULTS A SUBSTRATE B g4 2 TEMPERATURE CONTROLLER 5 663 638 Sheet 1 of 5 Sep 2 1997 U S Patent T Old C 8313038 5111538 SNO 3 1 149 318 S1I09419 IN JNIH YW NOILIN20238 51710534 33 19 3 393 10201084 c YOLVYINID WNOIS YOLVYVdIS YINIEGNOD WNOIS 9l 017 1 1 9 WNOIS 5 663 638 Sheet 2 of 5 Sep 2 1997 U S Patent o 96 9 0 3 31nGON S 130538 1N3WJN0SV3W M N3M0d 015 i 677 NI Y3MOd MOIS WALOV DL NI 43M0d LSV4 1 34014 31000N 1N3NW33f SV IN YIMOd 41SV 4 06 86 4311081N02 39NV4 OLAV 1804 1531 zz 66 5 663 638 Sheet 3 of 5 Sep 2 1997 U S Patent 0 8111081402 2 EL az 78 317315915 8 19 20 NI vL Y3IMOd v1130 62 W3MOd MOIS 135340 cc NOSIVdWOO 7717 1H911S 333308 m 95 x 28 93448 TAN Zl i UMMET ee T ENS uni 301V3931NI L E 000 A A DANG 1 15905 2Q r2 Jae M s Q Le end E a cc Sy ry 58 28 ge Uos I er NW v SNONOYHONAS i v 2 17062. should read are line 39 16dbm should read 16dBm and 8dbm should read 8dBm Signed and Sealed this Eighth Day of February 2000 Q TODD DICKINSON Arttesting Officer Commissioner of Patents and Trademarks
3. calorimetric bridge is to first heat the bridge externally and control the resulting temperature An actual insulated oven could be used if desired although it may be sufficient to rely upon a heated thermal mass without insulation A second solution to the problem of differential heating is to heat the bridge internally by an amount that varies according to the applied power to be measured This acts to eliminate the principal source of thermal gradients in the first instance By applying this heat as a common mode input signal to the two terminating loads of the calorimetric bridge the bridge itself does not become unbalanced and therefore does not attempt to measure that common power Actual input power to be measured does initially unbalance the bridge and is still measured as before A constant power circuit is response to the indication of measured power and acts to keep the sum of measured power plus the equal amount of feedback power that balances the bridge and the common mode power at a constant value corresponding to a full scale measurement or perhaps even a little more This eliminates internal thermal transients within the bridge that lengthen response time A solution to the problem of being unable to operate the integrator output sufficiently close to zero is to periodically remove the applied input and introduce into the servo loop a suitable slight offset from the true balance condition and then exactly balance the bridge
4. minus any residual measured power when power is not applied Of course the residual measured power does not represent real power but drift or other imbalance in the system One of these imbalances concerns the operating point of the main servo loop when there is indeed no power applied Let s assume for a moment that the system is truly ideal Then for zero RF in signals 40 and 41 are exactly equal and the error signal all around the loop would be zero In particular the output of integrator 52 would be zero Now if there were just a slight amount of power applied to the ACTUAL SLOW POWER IN 30 the error signal from error amplifier 42 would be rectified and produce say a positive signal to apply to integrator 52 In turn it would produce a positive going ramp that would eventually bring the loop back into balance If the integrator overshot or the actual input power level decreased the synchronous detector 45 would reverse the sign of the rectified error signal and the integrator would ramp back down The sign of the integrator output would still be positive however Now in a real system there are always imbalances and offsets however small This means that the loop will come into balance with a slight integrator output Absent the need to measure applied power that quiescent state of slight integrator output could as easily be a slight negative output after all that would still produce power in resistor 33 that ought to balan
5. receivers receiver 41 13 is coupled to various measurement circuits that provide mea sured data called RECEIVER RESULTS 15 The operation of the receivers 13 14 and signal generators 10 11 is controlled by a protocol generation and recognition machine 17 This is principally an embedded microprocessor con trolled system that interacts on the one hand with an operator through a display and control panel not shown or with external controller such as another computer system not Shown to select or specify the type of measurement or test to be performed and on the other hand with the internal resources of the test set e g the signal generators 10 amp 11 and receivers 13 amp 14 to set up and perform those mea surements and tests An important task of the protocol generation and recognition machine 17 is the ability to take into account the specific properties and behaviors that make up the various standards by which the radio communication equipment is to operate such as CDMA and GSM That said it will be appreciated that a signal combiner separator 3 couples the DUT 2 to the various signal gen erators and receivers in the test set 1 Principally the combiner separator 3 functions as a specialized power split ter and power combiner to allow simultaneous access to the DUT 2 by the signal generators 10 and 11 and by the receivers 13 and 14 This action is represented schematically by individual components 5 6 and 7 We shan t h
6. removal then a significant amount of time would need to pass to allow the thermal transient to expire before an accurate zero could be set Likewise after the zero set operation the re application of power would cause another thermal transient whose pres ence interferes with accurate readings The constant power circuit 60 eliminates these transients and allows the overall measurement system to dispense with thermal equilibration time associated with temporary imbalances introduced by legitimate changes in operating temperature It does this by removing those changes in operating temperature Finally if the constant power circuit is present a further purpose for either buffer amplifier 56 or multiplier 82 will be appreciated That further purpose is isolation That is they prevent any levels generated by the constant power circuit 60 from being measured by the ADC 84 and contributing incorrectly as it would be to the signal 27b and POWER MEASUREMENT RESULTS 9 I claim 1 An RF power meter comprising a calorimetric bridge including an incident power port coupled to receive an RF work signal whose power level is to be measured a comparison port coupled to receive a balancing signal a bias input coupled to receive a bias signal and first and second difference outputs at which appear respective first and second 5 663 638 1 difference signals derived from the bias signal and whose amplitudes differ in proportion to the differenc
7. temperature controller 61 is provided in conjunction with a heater resistor 39 and a sensing resistor 38 These two elements are not actually part of the bridge substrate 31 although they could be It will be recalled that we used an existing part and the keeper of the masks was cool to the 10 15 35 45 55 65 8 idea of modifying the part So we located the sensing resistor close to the bridge sensor put then both under a common canopy of thermal mass and then surrounded all of that with a ring of heater resistors All that stuff is mounted on a printed circuit board This controls the temperature of the calorimetric bridge sensor substrate to about one degree Celsius The whole works just described occupies about one half a square inch of board space for perhaps a quarter of an inch high Greater control of the temperature of the bridge sensor could be achieved if a genuine insulated oven were used Even then however there would still be short term changes in the average temperature of the various elements within the bridge owing to the application and removal of ACTUAL SLOW POWER IN 30 accompanied by the duplicate done by the main servo loop application and removal of COMPARISON SLOW POWER IN 57 These transient changes in operating temperature can reveal track ing errors in the temperature coefficients of the resistors making up the voltage dividers How can we keep the bridge 31 warmed as if power were applied
8. whose heat is coupled to another arm of the temperature sensitive bridge The applied power is measured by metering how much power is required of the servo loop to rebalance the bridge Calorimetric RF power meters of this sort have been known for some time See for example the Operating and Service Manual for the Hewlett Packard Model 434A Calorimetric Power Meter circa 1961 A more recent example of this technique can be found in an article on page 26 of the July 1987 Hewlett Packard Journal entitled Microprocessor Enhanced Performance in an Analog Power Meter The accuracy of calorimetric bridges is affected by temperature and especially by temperature differentials occurring across the bridge Even though the electrical value of the components within the bridge can be trimmed into balance they still have temperature coefficients and it is exceedingly difficult to control the thermal paths within the physical part The result is that under a thermal gradient such as the application of power to be measured the bridge can become unbalanced owing to an unsymmetrical response of the bridge itself to the thermal gradient In time if steady state conditions are maintained the gradient will level out and an accurate answer will be available It would 10 15 25 35 45 55 65 2 be desirable if this need for thermal time constant response time could be eliminated A control loop can use an amplifier or an integrator t
9. 21 may also be set to position 23 which couples the DUT to a fast power measurement module 25 that uses a diode power sensor The response time of the fast power measurement module 25 is in the microseconds and its output information 27a is made available as a part of POWER MEASUREMENT RESULTS 9 as well as being applied as an input to an auto range controller 28 The output of the auto range controller 28 is used to set the gain of the gain stage 20 The embedded control system periodically sets switch 21 to position 23 expressly so that the gain of gain stage 20 may be set to a value that places the output power level of the gain stage within the dynamic range of the slow power measurement module 26 It will be understood of course that the fast power measurement module 25 can also be used to perform primary power measurements on the DUT whenever a fast response is required e g pulsed power measurements Those interested in more information about what is in the fast power measurement module 25 may find it instructive to consult the aforementioned patent application Ser No 08 548 067 To continue we shall find it convenient to dwell on the arrangement that obtains when switch 21 is set to position 24 In such a case we shall say that the signal ACTUAL SLOW POWER 30 is coupled to the input of a slow power measurement module 26 whose power sensor is a calori metric bridge The output of the slow power measurement mo
10. 6 3114 1 19 90 NV E ovg 80 3 T 8 LY jp 0S JONVWWE OHVIS 3Nlj 310 INVLSNOO 8S 000 08 NI 43MOd 15 U S Patent Sep 2 1997 Sheet 4 of 5 5 663 638 63 FULL SCALE COMPARISON SLOW POWER IN FIG 4 107 NI 43MOd MOIS WALOV US Patent Sep 2 1997 Sheet 5 of 5 5 663 638 77 E nc Ld lt te HET gt V COMPARISON SLOW POWER IN 57 FIG 5A 77 e a be La a gt V COMPARISON SLOW POWER IN 57 FIG 5B COMPARISON SLOW POWER IN DELTA POWER 57 60 FIG 5C 5 663 635 1 CALORIMETRIC RF POWER METER WITH DYNAMIC ZEROING AND CONSTANT TEMPERATURE AND POWER DISSIPATION IN THE CALORIMETRIC BRIDGE REFERENCE TO RELATED APPLICATION The subject matter of this application is related to what is disclosed in U S Pat application Ser No 08 548 067 entitled METHOD AND APPARATUS FOR MEASURING RF POWER IN A TEST SET filed on 25 Oct 1995 pending by Melvin D Humpherys and assigned to Hewlett Packard Co U S patent application Ser No 08 548 067 is hereby expressly incorporated herein by reference BACKGROUND OF THE INVENTION Developments in semiconductors in the last decade or so and recent changes in the allocation of portions of the radio frequency spectrum have contributed to the vigorous growth i
11. US005663638A United States Patent us Patent Number 5 663 638 Humpherys 45 Date of Patent Sep 2 1997 54 CALORIMETRIC RF POWER METER WITH 57 ABSTRACT 75 73 21 22 51 52 58 56 DYNAMIC ZEROING AND CONSTANT TEMPERATURE AND POWER DISSIPATION IN THE CALORIMETRIC BRIDGE Inventor Melvin D Humpherys Spokane Wash Assignee Hewlett Packard Company Palo Alto Calif Appl No 681 564 Filed Jul 29 1996 GOIR 21 02 GOIR 19 03 324 95 324 106 324 95 96 105 3241106 455 61 1 67 1 226 1 226 4 115 References Cited U S PATENT DOCUMENTS 3 665 322 5 1972 Julie 3 845 388 10 1974 Ley et al 3 928 800 12 1975 Strenglein Primary Examiner Ernest F Karlsen Assistant Examiner Anh Phung The effects of differential heating inside the calorimetric bridge are reduced by first heating the bridge externally and controlling the resulting temperature The bridge is also heated internally by an amount that varies according to the applied power to be measured This acts to eliminate the principal source of thermal gradients in the first instance By applying this heat as a common mode input signal to the two terminating loads of the calorimetric bridge the bridge itself does not become unbalanced and therefore does not attempt to measure that common power Actual input power to be measured does initi
12. a digital compen sation value and also comprises a digital to analog converter 10 15 25 30 12 having an input coupled to the digital compensation value and having an output coupled to the compensation output 5 An RF power meter as in claim 1 further comprising a heater resistance thermally coupled to the calorimetric bridge a temperature sensor thermally coupled to the calorimetric bridge and a temperature controller coupled to the temperature sensor that maintains the calorimetric bridge at a constant ambient temperature 6 An RF power meter as in claim 1 wherein the servo circuit includes an integrator and the RF power meter further comprises a circuit in series with the RF work signal before the RF work signal is applied to the incident power port that at selected times reduces the power level of the RF work signal at the incident power port to substantially zero an offset circuit that during the selected times causes the servo circuit to produce for the balancing signal a selected fixed value corresponding to a minimally detectable non zero power level for the RF work signal and a zero balance circuit coupled to an error signal within the servo circuit that during the selected times nulls to zero that error signal by varying the conduction in a shunt between the bias input and the first difference output and that during other times maintains the conduction in the shunt at its most recent value during a s
13. ally unbalance the bridge and is still measured as usual A constant power circuit is response to the indication of measured power and acts to keep the sum of measured power plus the equal amount of feedback power that balances the bridge and the common mode power at a constant value corresponding to a full scale measurement This eliminates internal thermal transients within the bridge that lengthen response time The feedback signal to balance the calorimetric bridge is produced by an integrator To allow operation of the integrator output closer to zero for measurement of small applied signals the applied input to be measured is periodically removed and a suitable slight offset from the true balance condition is introduced into the servo loop The bridge is then exactly balanced with a separate sample and hold control loop Periodically can mean once a second ten times a second or as is preferred at the start of each measurement Attorney Agent or Firm Edward L Miller 6 Claims 5 Drawing Sheets p zl STATIC BALANCE 50 ACTUAL C AA 47 387 T PENER SE T INTEGRATOR thm 1 32 a on MULTIPLIER 733 49 ae w x LOW PASS LOG SYNCHRONOUS L d FILTER AMP DETECTOR 3 1 ree 42 1 502 39 82 43 383 44 45 3 E i CALORIMETRIC 4 58 ven Spr 37 OCA BRIDGE E k 25 kHz 52 ww v SUBSTRATE F1 X ke 1 ee nae ern ge Bs
14. ave much more to say about the elements 5 6 and 7 in combiner separator 3 save that they must provide certain desirable isolation between various ones of the DUT and the signal generators and receivers must not be lossy and must be fairly broad band say flat from below 50 MHz to above 2 GHz Our present interest is in the measurement of RF power present in the path 18 connecting the combiner splitter element 5 to the DUT 2 To this end note sampler 4 which may be a suitable directional coupler which extracts a signal say 20 db below that present in line 18 and applies it to a power measurement module 8 whose output may be called POWER MEASURE MENT RESULTS 9 POWER MEASUREMENT RESULTS 9 may be taken with RECEIVER RESULTS 15 as comprising the raw measurements of the performance of the DUT 2 These measurements are interpreted by the protocol generation and recognition machine 17 Refer now to FIG 2 wherein is shown in block diagram form 19 an expansion of the power measurement module 8 5 10 15 25 35 45 55 65 4 RF power from the DUT 2 is coupled by line 18 and coupler 4 to a gain stage 20 whose gain may be set to a convenient amount which includes zero The signal from the gain stage 20 is applied to a switch 21 When the switch 21 is thrown to position 22 the output of the DUT is via the gain stage 20 connected to a test port This facilitates test and cali bration of the test set itself The switch
15. ce the loop Heat is heat Ah but what then of the logical sense of the control signal If the loop overshoots the polarity of the signal from the synchronous detector 45 would still be negative which would drive the output of the integrator 52 still more negative further adding to the overshoot In other words the logical sense of the control signal has been reversed The result is that the servo drives itself to the rail and locks up Clearly this mode of operation must be avoided As a practical matter this means that actual operational zero point of the loop must not be set so close to the actual zero crossover of the integrator that noise drift or other loss of margin conditions inadvertently pushes the servo system over the edge Absent some better mechanism the brute force accommodation for this situation is to simply put the operational zero point at what is really the ten percent point in the safe operating region This works but uses up dynamic range and make the measurement of low power levels less accurate Reference to FIG 4 will be useful at this point FIG 4 is a graph 62 of the COMPARISON SLOW POWER IN along abscissa 64 needed to null the servo loop for different mounts of ACTUAL SLOW POWER IN along ordinate 65 Because of the symmetry of the calorimetric bridge 31 the slope of the resulting line in the graph is one ten milliwatts of comparison signal 57 balances ten milliwatts of input signal 30 Thus line 63 is de
16. d for measurement then signal 57 is at its minimum the dynamic zero loop keeps it close to but safely away from zero remember then the constant power circuit 60 needs to produce exactly equal full power heating in each of resistors 32 and 33 If the input power to be measured were half of full scale then the balance of full scale would be made up in the terminating resistors 32 and 33 by the constant power circuit 60 And if the input power to be measured were full scale then no power would be supplied by the constant power circuit 60 So the constant power circuit 60 simply maps the mini mum to maximum excursion of signal 57 into an appropriate excursion at the common end of resistors 58 and 59 The exact relationship of that mapping is determined by chosen resistance of 400Q for resistors 58 and 59 in comparison to the values of 50Q for resistors 32 and 33 The value of 4000 for resistors 58 and 59 was selected to minimize the dis continuity caused by their presence Despite what was earlier said about the possibility that COMPARISON SLOW POWER IN 57 could be either an 5 663 638 9 AC or a DC signal it can now be appreciated that if the constant power circuit 60 is to be employed it may be advantageous if COMPARISON SLOW POWER IN 57 is an AC signal In this way it can be arranged that there be DC blocks at locations 72 and 73 In this way it is now certain that the DC power added by the constant power circuit will indeed be di
17. dule 26 is information 27b that is part of the POWER MEASUREMENT RESULTS 9 Our interest is primarily in certain aspects of the operation and construction of the slow power measurement module 26 Now consider the block diagram shown in FIG 3 What is shown is a simplified block diagram of the slow power measurement module 26 At the upper left of the block diagram is the input signal ACTUAL SLOW POWER IN 30 while at the lower right appears the output signal 27b that is part of POWER MEASUREMENT RESULTS 9 We are interested now in what goes on inbetween these two signals The input signal ACTUAL SLOW POWER IN 30 is applied to a calorimetric bridge substrate 31 that includes a 500 terminating resistor 32 that is closely thermally coupled to a sensing resistor 35 whose resistive value has a signifi cant dependence upon temperature Sensing resistor 32 is in series with a load resistor 34 and forms a voltage divider therewith A reference signal generator 46 produces a 5 KHz signal 47 that among a number of other places is applied to the top of the voltage divider formed by resistors 34 and 35 Thus the amplitude of the 5 KHz signal 46 at the tap of the voltage divider is principally determined by the amount of power being dissipated in terminating resistor 32 Now observe the signal COMPARISON SLOW POWER IN 57 It is a DC or low frequency signal applied to a 500 resistor 33 identical to resistor 32 Resistor 33 is closely thermal
18. e in power level between the work signal and the bal ancing signal a servo circuit coupled to the first and second difference 5 signals and producing therefrom the balancing signal the servo circuit adjusting the power level of the balancing signal to minimize the amplitude difference between the first and second difference signals a compensation circuit having an input coupled to the balancing signal and having a compensation output first and second isolation networks each coupled at one end to the compensation output and respectively at their other ends to the incident power port and the compari son port and the compensation circuit applying equal amounts of addi tional power to the incident power port and to the comparison port the compensation circuit adjusting these additional powers such that the sum of these additional powers the power of the work signal and the power of balancing signal is a constant 2 An RF power meter as in claim 1 wherein the first and second isolation networks are resistors whose values of resistance are equal 3 An RF power meter as in claim 1 wherein the com pensation circuit comprises an analog shaping amplifier 4 An RF power meter as in claim 1 further comprising an analog to digital converter coupled to the balancing signal to produce a digital power level signal and further wherein the compensation circuit comprises a look up table addressed by the digital power level signal to produce
19. elected time k X UNITED STATES PATENT AND TRADEMARK OFFICE CERTIFICATE OF CORRECTION Page 1 o 2 PATENTNO 5 663 638 age of DATED September 2 1997 INVENTOR S Melvin D Humpherys It is certified that error appears in the above identified patent and that said Letters Patent is hereby corrected as shown below Column 2 line 16 not should read no and line 46 response should read responsive Column 3 line 1 delete a after is line 8 insert a after Fig 4 lines 29 and 30 between with and external insert an line 58 20db should read 20dB Column 4 line 31 SLOWPOWER 30 should read SLOW P OWER IN 30 line 65 occupied should read coupled Column 5 line 1 signal should read signal 40 line 3 that ACTUAL should read that of ACTUAL and line 4 signal 30 should read signals 30 Column 7 line 11 fight should read right line 25 Case should read case line 29 at say should read at say and line 48 hulling should read nulling Column 8 line 61 insert after by UNITED STATES PATENT AND TRADEMARK OFFICE CERTIFICATE OF CORRECTION Page 2 of 2 PATENT NO 5 663 638 DATED September 2 1997 INVENTOR S Melvin D Humpherys It is certified that error appears in the above identified patent and that said Letters Patent is hereby corrected as shown below Column 9 line 36 is
20. even when it isn t Suppose that no power is being applied via lines 30 and 57 Now apply duplicate amounts of power via resistors 58 and 59 to resistors 32 and 33 respectively Let the amount applied be approximately the amount resistors 32 and 33 would experience under full scale measurement conditions An actual measurement initially unbalances the bridge and so provokes a rebalancing via the error signal However these conditions we speak of now are not measurement conditions We are assuming that we apply the same power to each of resistors 32 and 33 so their heating does not unbalance the bridge no error signal appears at error ampli fier 42 and integrator 52 does not ramp MEASUREMENT CONDITIONS DO PRODUCE AN INITIAL UNBAL ANCE OF THE BRIDGE WHILE DRIVING RESISTORS 58 AND 59 DOES NOT Resistors 58 and 59 are driven by a signal DELTA POWER 74 that is produced by a constant power circuit 60 The main servo loop cannot tell if constant power circuit 60 is driving resistors 58 and 59 or not for its effect is common mode and does not unbalance the calori metric bridge 31 So the main servo does its thing for applied power regardless of whether or not the constant power circuit is present The constant power circuit 61 knows how much power DELTA POWER 74 to apply to the bridge via resistors 58 and 59 because it monitors the same signal 57 that is metered to determine the POWER MEASUREMENT RESULTS 9 If zero power is being applie
21. haps other amplifiers One of the reasons that such an approximation works quite well is that at low power levels the exact graph 75 in FIG 5A has the lowest dx dy and at high power levels has the greatest dx dy These extremes are perhaps the regions of greatest interest and are fairly well represented by the approximation produced by the shaped amplifier circuit of FIG 5C If the constant power circuit 60 were implemented with a DAC not shown driven by a look up table or computation based on signal 27b then the approximation could be very near to perfect indeed Now what of those who desire that there be a value of resistance at location 81 that is different from that of the isolation resistors 58 and 59 Perhaps there is to be no resistance at all at that location What then Well all that changes is that the graph 75 in FIG 5A changes from being a quadrant of a circle to a quadrant of some ellipse It is no more difficult to approximate that with either a circuit such as shown in FIG 5C or with a DAC driven by a look up table or computation based on signal 278 than it is to do it for a quadrant of a circle It will be appreciated that the constant power circuit 60 enhances the operation of the dynamic zero mechanism 49 50 71 Why this is so may be understood by recalling that the dynamic zero mechanism first removes the applied input 30 and then does the zero If there were a substantial level of power applied at the time of
22. ly coupled to a sensing resistor 37 identical to sensing resistor 35 Resistor 37 is also in series with a load resistor 36 identical with load resistor 34 so that resistors 36 and 37 form a voltage divider corresponding to the voltage divider formed by resistors 34 and 35 The voltage divider formed by resistors 36 and 37 is also occupied at its top to the 5 KHz reference signal 47 Ideally the amplitude of the signal 41 produced at the tap of this second voltage divider 5 663 638 5 will equal that of the signal whenever power level of the signal COMPARISON SLOW POWER IN 57 is the same as that ACTUAL SLOW POWER IN 30 It is the job of the circuitry inbetween the two signal 30 and 57 to servo signal 57 to match signal 30 by nulling to zero the difference between signals 40 and 41 That done the power level of signal 30 can be inferred by a direct measurement of the amplitude of signal 57 and the knowledge that resistor 33 is 50 Along the way however there arise some significant opportunities for error To continue the calorimetric bridge substrate 31 is a unified assembly fabricated using thin film and integrated circuit techniques The 50Q resistors 32 and 33 are formed of meandering titanium traces for example and the entire assembly 31 is packaged up in its own little can It is not however a part that was specifically designed for this particular application It was designed to be an RF power sensor to be sure but for us hap
23. n the uses and markets for various portable transceivers such as used in cellular telephone service and GMRS That which is to be manufactured must also be tested and as needed repaired and tested The advent of spread spectrum technologies such as CDMA in the United States and GSM in Europe require that additional levels of sophistication be included in the test equipment to allow that aspect of the performance of the device under test to be evaluated Such test equipment ought to be reliable and relatively compact those wishing to sell systems comprising a rack full of separate instruments need not apply as well as adaptable to future developments Among the things that such a test set has to do is measure RF power The power measurement module of such a test set ought therefore to be small accurate broad band suitable for pulse modulation applications in addition to CW and inexpensive That is quite a list of disparate requirements and poses a significant challenge to the designers of a test set to be used with radio equipment such as cellular telephones A calorimetric RF power meter is a device that accepts RF power into a terminating load and thermally couples the heat generated to a temperature dependent resistance that is one arm of a bridge This unbalances the bridge and pro duces an error signal within a servo loop The servo responds by applying DC or low frequency power to a separate but identical terminating load
24. nd 59 in order to get it to the terminating resistors 32 and 33 For the sake of simplicity let us also temporarily assume that buffer amplifier 56 is absent and that in its place is multiplier 82 and 25 KHz reference signal 83 Now the range of power levels that may be applied to either of the terminating resistors 32 and 33 is from 16 dbm to 8 dbm or roughly from 25 mv to 500 mv for a 50Q system The analog multiplier 82 is much better suited to operating at a higher level several volts The solution is to attenuate the output of the analog multiplier 82 rather that try to operate it at levels where noise and nonlinearity are less favorable Suppose we put a 4000 resistor at location 81 this produces a 9 1 attenuation between the output of the analog multiplier 82 and what is produced across terminating resistor 33 Note that it is now also necessary to adjust the interpretation of the output 27b of the ADC 84 it is now nine times too big Overall this is good for the multiplier performance reasons stated above but it also provides the simplicity or removes a complication that was alleged in the first sen tence of the paragraph The simplicity arises because now the circuits 82 or 56 that originate COMPARISON SLOW POWER IN 57 and the circuit 60 that originates DELTA POWER 74 all involve the same nine to one attenuation Because of this symmetry our earlier statements about Pop tPcomptAP C and 2P om tAP C can be understood not
25. nder the control of another control loop With reference again to FIG 3 here is how it is done First we need a way to counter any imbalance in the calorimetric bridge 31 traceable to the two resistive dividers 34 35 and 36 37 The first tool for this is the ability to remove any input otherwise present on line 30 ACTUAL SLOW POWER IN This is accomplished by setting the gain of gain stage 20 see FIG 2 to zero The second tool for this is resistor 70 and DAC 51 think programmable resistor Resistor 70 is small enough to by itself ensure that the loop gets unbalanced one way no matter where in the allowable range of tolerances the voltage dividers end up DAC 51 is then chosen to be able to move the loop back the other way by at least an amount equal to any worst Case imbalance in the first way This is a long way of saying that it is guaranteed that some setting for DAC 51 will serve to balance the loop right at the true zero crossover perhaps plus even a little more for safety DAC 51 is initially set at say the one percent mark Note that DAC 51 is not an active part of the dynamic zero control loop promised above it simply provides an initial operating point which may drift but nevertheless centers the range of that dynamic zero con trol loop The control loop we seek dynamically sets the operational zero point of the main servo loop To do this switches 48 and 55 are periodically set together to the positions
26. ng an instance of dynamic zeroing or to an integrator 52 at other times Integrator 52 responds to the continued presence of a rectified error signal by ramping its output up or down until the rectified error signal vanishes During these other times i e during power measurement the output of the integrator 52 is coupled via switch 55 to a buffer amplifier 56 which may be a unity gain amplifier It supplies the heft needed to drive the relatively low impedance of the 50Q resistor 33 It will further be appreciated that buffer amplifier 56 might be replaced by an analog multiplier circuit 82 whose two inputs were the output of the integrator 52 via switch 55 and a low frequency AC signal 83 say 25 KHz In this way the signal COMPARISON SLOW POWER IN 57 would become an AC signal instead of one that is DC Once this main servo loop comes into balance which is pethaps on the order of several milliseconds an ADC analog to digital converter 84 measures the amplitude of the integrator output signal applied to the input of the buffer amplifier 56 or to the input of multiplier 82 to produce output 27b which is part of POWER MEASUREMENT RESULTS 9 We now consider certain improvements that make the thing work better To begin with it will be appreciated that a power measurement is really a difference measurement 10 15 35 40 45 55 65 6 That is what we really seek is the measured power when power is applied
27. o form the feedback signal In many situations where a small change is to be measured and a DC measurement technique is prone to unacceptable drift it is advisable to shift to AC measurement techniques In these cases a synchronous detector coupled to an integrator form an attractive combi nation for forming the actual feedback signal from the amplified error signal If as in the case of a calorimetric power measurement technique the integrator output is used to produce heat then either polarity of output from the integrator can cause a point of balance However one polarity has associated therewith the wrong logical sense of change as between the feedback signal and the error signal If the system gets into that state the feedback is not longer negative but becomes positive As a result it is generally necessary to anticipate a worst case of margins and then prevent the integrator output from getting within that close to zero even when the bridge is actually balanced As a complication to the logic of the servo loop that brute force solution is indeed manageable but it can nevertheless have the disadvantage of limiting the accuracy with which small signal levels can be measured It would be desirable if this limit on dynamic range could be eliminated by allowing the integrator output to operate much closer to zero when the bridge is indeed balanced SUMMARY OF THE INVENTION A solution to the problem of differential heating inside the
28. only as applying to the 500 terminating resistors 32 and 33 but also to the series combination of those with an appro priate one of the 400Q isolation attenuation resistors With all these preliminaries safely understood now con sider the graph 75 in FIG 5A The abscissa 76 represents the voltage produced at the output of analog multiplier 82 or possibly at the output of buffer amplifier 56 so long as there is a 400Q resistor at location 81 The ordinate 77 represents the corresponding voltage that is to be produced for DELTA 10 15 35 45 55 60 65 10 POWER 74 by the constant power circuit 60 It can be shown given the simplifications set out above that the curve 75 is a quadrant of a circle Analog circuits that have exactly such a transfer func tion from input to output are well difficult It turns out that in the present situation an approximation works quite satis factorily The approximation is shown in FIG 5B as the graph 78 It is a piecewise linear approximation that can be obtained by those skilled in the art using any of a variety of shaped amplifier circuits similar to what is depicted in FIG 5C FIG 5C shows the generalized form a shaped amplifier circuit that may be used as an analog instance of constant power circuit 60 As seen in the figure the basic idea is to surround an amplifier 79 with a nonlinear feedback network 80 The nonlinear feedback network 80 may contain diodes zeners and per
29. oppo site the way they are shown in the figure This connects the output of the synchronous detector 45 to the input of integrator 49 whose output 50 is applied as one input to an analog multiplier 71 The other input to the analog multiplier 71 is the 5 KHz reference signal 47 The output of the analog multiplier 71 may be thought of as a false exror signal as if from the voltage dividers in the bridge 31 that is coupled to the error amplifier 42 As before the main servo system nulls itself so that the output from the error amplifier goes to zero Now if there were no power going into bridge resistor 33 during this operation this hulling would indeed tend to put the main loop right onto the exact loop crossover point a very risky thing to do and almost certain to result in disaster However we do not let the power in bridge resistor 33 be zero No Sir we set it at the one percent mark by the action of switch 55 and the slight offset 54 applied to buffer amplifier 53 Thus it is that the dynamic zero servo loop of integrator 49 and analog multiplier 71 nulls to the one percent mark Once this null has been achieved switches 48 and 55 are set back to their normal positions Integrator 49 simply holds its last output with very little drift until the next dynamic zero cycle A further enhancement in the performance of the calori metric bridge 31 may be obtained by arranging that it operate at a constant temperature To this end a substrate
30. pens to be an already existing part whose specifications and tolerances were selected to suit an earlier application For reasons of economy it is desirable to take advantage of this existing part and find a way to live with certain shortcomings that result It is not that the part is a bad part it is more that if we were designing it for the first time we would change some things to be more appropriate to our intended use Even if we did however issues we are about to explore do not altogether vanish So let us first describe the overall servo loop that nulls the difference between signals 40 and 41 by adjusting the power level of signal 57 That done we can return to a description of the shortcomings and what to do about them To begin then note that signals 40 and 41 are coupled to a difference amplifier 42 whose output is applied to a low pass filter 43 The main purposes of the low pass filter 43 are to assist in setting the dynamic behavior step response of the servo loop and to suppress harmonics of the amplified error signal The output of the low pass filter 43 is coupled to a log amplifier 44 whose output is in turn applied to a synchronous detector 45 The synchronous detector 45 oper ates to rectify the amplified 5 KHz error signal the resulting DC signal has a polarity indicative of whether COMPARI SON SLOW POWER IN 57 is too high or too low and is applied via switch 48 to either an integrator 49 duri
31. picted as being at forty five degrees to the axes Range 67 would be the ideal range over which the loop would operate if there were no concerns about the output of the integrator 52 going negative However as we have seen that is not a practical realization dotted line 66 represents the unhealthy case To keep the zero applied power condition from approaching the actual crossover point of the loop one might adopt operating range 68 This would be done by simply building into the loop sufficient offset to prevent the worst case pile up of bad margins from producing an actual operating point that was too close the origin of the graph 62 One simply then scales the range 68 to represent the range of actual applied power That is the ten percent point at the lower end of range 68 represents no applied power while the full scale point 5 663 638 7 represents full scale applied power As mentioned above it is not that this does not work It is more that it interferes with dynamic range and makes measurement of low power levels less accurate A better way would be to be able to safely adopt range 69 as the operating range it s lower end is at say one percent This is desired even though we are either unable to or choose to not decrease the offsets and imbalances that combine to suggest the need for a ten percent margin in the first place Simply put we instead choose to operate the main loop fight next to the hairy edge as it were u
32. ssipated in the terminating resistor 33 regard less of the presence or absence of COMPARISON SLOW POWER IN 57 To the same end it could be arranged that the power provided by the constant power circuit 60 be delivered as AC The particular way we have shown the constant power circuit 60 is one that is perhaps best suited for an analog circuit that operates on the analog value of COMPARISON SLOW POWER IN 57 to produce the drive DELTA POWER 74 to resistors 58 and 59 An alternate method of producing the drive to resistors 58 and 59 is to let the constant power circuit 60 include a DAC whose setting is obtained either from a calculation or a look up table based on the digital value 275 POWER MEASUREMENT RESULTS 9 Certain aspects of one possible such constant power circuit 60 is shown in FIGS 5A C The relationship we are interested in is FAP C By this we mean that the applied power to be measured 30 and the resulting com parison power 57 produced to balance the calorimetric bridge when added to delta power 74 sum to some constant amount probably best taken to be twice full scale for Papp We could also say that 2 For the sake of brevity we omit the rigorous demonstration merely reminding the reader to keep in mind that power is propor tional to the square of the voltage and that overall amount of power that the constant power circuit 6 needs to produce must be divided among the isolation resistors 58 a
33. with a separate sample and hold control loop Periodically can mean once a second ten times a second or as is preferred at the start of each measurement BRIEF DESCRIPTION OF THE DRAWINGS FIG 1 is a simplified block diagram of a test set for radio communication equipment such as cellular telephones and GMRS transceivers and that includes an RF power mea surement capability constructed and operating in accordance with the invention 5 663 638 3 FIG 2 is a an expanded yet still simplified block diagram of an RF power measurement portion of the test set FIG 1 FIG 3 is an expanded block diagram for a portion of the block diagram of FIG 2 representing a calorimetric RF power meter that includes dynamic zeroing and constant temperature and power dissipation in the calorimetric bridge FIG 4 is diagram indicating certain aspects of operation that are of interest for the calorimetric RF power meter of FIG 3 and FIGS 5A C are diagrams describing a constant power circuit shown in the block diagram of FIG 3 DESCRIPTION OF A PREFERRED EMBODIMENT Refer now to FIG 1 wherein is shown a simplified block diagram 1 of a test set for use with radio communication equipment such as cellular telephones and GMRS trans ceivers A DUT 2 or Device Under Test represents the radio communication equipment to be tested In this particular test set 1 there are two signal generators 10 and 11 and two receivers 13 and 14 Of the two
Download Pdf Manuals
Related Search