|
|
On Mon, 06 Dec 2004 01:05:54 GMT, Richard Clark
wrote: On Sun, 05 Dec 2004 22:51:03 GMT, (Robert Lay W9DMK) wrote: New Measurements - I created a terminating load consisting of 4 composition resistors in parallel. That measured 4.3 + j0.65 AT 20 MHz. I threw together two Allen Bradley 10 Ohm 5% 1/4 Watt resistors and came up with 5.1 -j0.5 in a quick test at 20 MHz. I then measured the input impedance of the 5.33 meter length of RG-8/U Foam coax terminated with the 4.3 +j0.65 ohm load at 20 MHz, and that was 5 - j7.1 ohms. The SWR at the load is 11.63 and the SWR at the input is 9.88. Using a velocity factor of 0.745 and an attenuation value of .77, I calculated the theoretical input impedance of the coax with the above terminator. That gave a result of 5.17 - j7.3 ohms (theoretical). The SWR at the load is 11.63, and the SWR at the input to the line is 9.88 (theoretical). In setting up the simulation, it is necessary to pick an attenuation and a velocity factor that are not only within the normal distribution for that particular coax but also give a reasonably good match with the measured values. In my opinion, the values that I used in the simulation are well within the normal distribution of values for this type of line, which has published values of VF=.8 and attenuation = 74 at 20 MHz. Hi Bob, I would say that your data shows a very good correlation to the models and certainly the presumptions you made are well within the production variables. The simulation also predicts the losses, and I used two different models for that calculation. Both loss models predict a total loss of 0.723 dB, which is 0.589 above the matched line losses based on the normal attenuation. The two math models used were as follows: 1) ITT Reference Data for Radio Engineers, 5th Edition, pages 22-8 and 22-9. 2) The ARRL Antenna Book, 17th Edition, page 24-9. Based on the limited tests that I have made so far, the two models seem to give the same results. However, I am hoping to be able to conduct measurements on configurations that involve much higher SWR values. The immediate problem to be overcome is the measurement of such impedance values as will be encountered. Measure Q by the BW of the Half Power points. 73's Richard Clark, KB7QHC Dear Richard, I finally created a test load that gives me the higher SWR that I wanted. It measures 7.0 - j2008 at 1.8 MHz. I placed that test load at the end of a 150 foot piece of RG-59/U and measured the input impedance as 38.5 + j 151.6 at 1.8 MHz. The load SWR is 7901 and the input SWR is 10.18. Solution of the transmission line equations for this particular load and with coax characteristics of 73 ohms, VF = .646 and an attenuation of 0.57 dB per 100 feet gives an input impedance of 38.67 + j 149, which is a very good match to the measured value. Losses are calculated using the same two methods as reported in my previous posting, as follows: 1) ITT Reference Data for Radio Engineers, 5th Edition, pages 22-8 and 22-9. 2) The ARRL Antenna Book, 17 th Edition, page 24-9. Matched line losses = 0.855 dB Additional losses = 28.087 dB Total losses = 28.942 dB I am satisfied that the methods of calculating losses as described in the two references are in agreement and are valid. I am also reasonably satisfied that the 1 dB steps that are printed on Smith Charts as the means of determining matched line losses are valid, as are the nomograms provided in the ITT Handbook, pages 22-7 and 22-8, above. 73, Bob, W9DMK, Dahlgren, VA http://www.qsl.net/w9dmk |
|
On Mon, 06 Dec 2004 19:57:34 GMT, Richard Clark
wrote: On Mon, 06 Dec 2004 18:03:47 GMT, (Robert Lay W9DMK) wrote: I finally created a test load that gives me the higher SWR that I wanted. It measures 7.0 - j2008 at 1.8 MHz. I placed that test load at the end of a 150 foot piece of RG-59/U and measured the input impedance as 38.5 + j 151.6 at 1.8 MHz. The load SWR is 7901 and the input SWR is 10.18. Solution of the transmission line equations for this particular load and with coax characteristics of 73 ohms, VF = .646 and an attenuation of 0.57 dB per 100 feet gives an input impedance of 38.67 + j 149, which is a very good match to the measured value. Losses are calculated using the same two methods as reported in my previous posting, as follows: 1) ITT Reference Data for Radio Engineers, 5th Edition, pages 22-8 and 22-9. 2) The ARRL Antenna Book, 17 th Edition, page 24-9. Matched line losses = 0.855 dB Additional losses = 28.087 dB Total losses = 28.942 dB I am satisfied that the methods of calculating losses as described in the two references are in agreement and are valid. I am also reasonably satisfied that the 1 dB steps that are printed on Smith Charts as the means of determining matched line losses are valid, as are the nomograms provided in the ITT Handbook, pages 22-7 and 22-8, above. Hi Bob, Congratulations are in order for your effort at the bench, regardless of outcome. Congratulations are in order for your chain of reasoning, your attention to detail, and the obvious refinement of technique that is now agreeing not only with references, but is also consistent from one test to the next. I still see some aberration in the data when you have to drive the cable Z to 73 Ohms to make the formulas work. It shows that the generality of the references is good, but with the instance of your data is a forced conclusion. I don't find that particularly upsetting as the accumulation of error could easily account for some of the differences. You might want to revisit some of Bart's offerings in this thread; especially his discussion of the effect of Low-R loads as a source of Hi-Z, Hi-SWR. 73's Richard Clark, KB7QHC Dear Richard, No - I didn't. That was a surprise to me, too! The specs on RG-59/U say 73 ohms. There are 2 other RG-59 types - RG-59 Foam (75 ohms) and RG-59A, also 73 ohms. C'est la Guerre! 73, Bob, W9DMK, Dahlgren, VA http://www.qsl.net/w9dmk |
|
On Mon, 06 Dec 2004 23:36:13 GMT, Richard Clark
wrote: On Mon, 06 Dec 2004 22:25:03 GMT, (Robert Lay W9DMK) wrote: No - I didn't. That was a surprise to me, too! The specs on RG-59/U say 73 ohms. There are 2 other RG-59 types - RG-59 Foam (75 ohms) and RG-59A, also 73 ohms. Hi Bob, You are, in the expression common here in the NW, whipsawing me with your changes in cable type. I failed to catch that in your post. Given you are using 73 Ohm line, your results are very remarkable. I'm glad you've gone the distance. This group, however, exhibits a very odd ethic that I would call the reverse Little Red Hen story. Through more than 150 postings they all want to offer advice on how to cut the grain; they will tell you where to mill it into flour; they will offer you recipes on what to bake; but none seem to be around to praise the chef or the cake. 73's Richard Clark, KB7QHC Dear Richard, Well, I only came here to share, so I'm not the least disappointed - as some say, the pleasure is in the doing! You might have asked why I changed coax. I didn't feel that it would be as dramatic with only 17.5 ft of line. So, I went to the barn and looked for the biggest, cleanest roll of coax hanging on the wall, and that was it - Hi! Bob, W9DMK, Dahlgren, VA http://www.qsl.net/w9dmk |
|
I never did quite get clear about your thesis. Would you mind restating it?
Roy Lewallen, W7EL Richard Clark wrote: Hi Bob, I didn't ask why because it was evident from the drama of results. May as well force the numbers to expose the theory. I tried that with my own thread when you started this one - it n'er came out as well as yours. However, your confirmation of the loss does support my thesis, even if my thread did not. 73's Richard Clark, KB7QHC |
On Mon, 06 Dec 2004 19:03:47 -0800, Roy Lewallen
wrote: I never did quite get clear about your thesis. Would you mind restating it? Source Z matters. |
"Richard Clark" wrote in message
... On Mon, 06 Dec 2004 19:03:47 -0800, Roy Lewallen wrote: I never did quite get clear about your thesis. Would you mind restating it? Source Z matters. What is the source Z of a solid state power amplifier, or even a tube amplifier? 73, Frank |
Richard Clark wrote:
On Mon, 06 Dec 2004 19:03:47 -0800, Roy Lewallen wrote: I never did quite get clear about your thesis. Would you mind restating it? Source Z matters. Guess I didn't misunderstand after all -- it actually was so vague as to be meaningless. Thanks for the elaboration. Roy Lewallen, W7EL |
"Richard Clark" wrote: Source Z matters. The *magnitudes* and *phases* of Vfor and Vref are affected by Zs as described by Chipman. But I don't find anything in Chipman's book to indicate that the *ratio* of Vmax to Vmin (VSWR) is affected by Zs. Perusing all the references to VSWR in Chipman's book, you will find that the source is not mentioned at all. Only the load reflection coefficient and the transmission line characteristics are needed to calculate VSWR. -- 73, Cecil http://www.qsl.net/w5dxp |
Perusing all the references to VSWR in Chipman's book, you will find that the source is not mentioned at all. Only the load reflection coefficient and the transmission line characteristics are needed to calculate VSWR. =================================== Cecil, the trouble with 'bibles' is that they are so easily misquoted. It's always better to rely only on your 'own' knowledge. Only the MAGNITUDE of the reflection coefficient is needed to calculate SWR. The phase angle is superfluous. Nothing else whatever need be known about the line. Not even Zo, the terminating impedance, and certainly not the generator impedance. Conversely, the SWR will tell you virtually nothing about what's going on on the line until you include and add to it what you already know anyway. You can't even work back to find the reflection coefficint because of the loss of the angle information. The reflection coefficient is of no use to anybody without its angle, except, of course to calculate the SWR. Abolish SWR meters! --- Reg. |
On Tue, 07 Dec 2004 13:02:01 GMT, "Frank"
wrote: What is the source Z of a solid state power amplifier HI Frank, Commonly 1.5 to 3 Ohms resistive transformed to 35 Ohms to 70 Ohms at the Connector. or even a tube amplifier? Much greater variation here, X KOhms resistive transformed to 50 Ohms at the connector. Such are ballpark figures, as an average over a full cycle, at rated power, for Class AB operation in a Push-Pull configuration. This typically results in an efficiency on the order of 40% to 60%. The nut of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then almost all their power would be reflected back into them at the antenna connector's connection to a 50 Ohm antenna system. The argument of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then reflections from the load (once you got some power into line) would encounter this same massive mismatch and re-reflect. There is a naive argument here (all too common) that this is "exactly" what happens. My pointed observation to those statements is "why would anyone need a tuner then?" The refinement of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then with any mismatch at the load you haven't got a clue what power is being applied OR reflected. This is called the Mismatch Uncertainty. It is another indicator of the failure of the argument mentioned in the previous paragraph. The relevance of the matter about "Source Z matters" here is that the references that Bob used to measure and model line loss supports this thesis. It presents an opportunity to observe how a line would suffer additional loss through being mismatched at both ends. In this regard, it would be due to the fictive argument for Source Z being very much lower or very much higher than 50 Ohms (in other words, lacking the transform circuitry commonly found in commercial gear). When the loss is not observed, the fiction is shown. 73's Richard Clark, KB7QHC |
Hi Richard,
With solid state power amplifier design; the criteria was always that you must present an impedance, to the output devices, such that the desired output power is delivered to the load (while not exceeding device dissipation). Any attempt to optimally match the load to the source impedance will result in over-dissipation, and probable destruction of the source device -- probably by excess collector/drain current. If you remember, Motorola used to publish Smith charts of the output impedance for their power amplifier devices. Talking to one of Motorola's design engineers; I asked "How do you derive these Charts". His answer was; "We use a matching network and adjust it for the required output power, then measure the input impedance of the network. The complex conjugate of this impedance is then defined as the source Z". The fact is these data are not the actual source Z of the device, but are probably considerable higher. I don't remember anybody actually trying to measure the large signal S parameters of solid state devices. I seem to remember that tube amplifiers were designed based on the source impedance calculated as 2Vp/Ip, (Where Vp is the plate voltage, and Ip the plate current), and have no idea how, or if, it relates to the actual source Z of the device. Anyway, I am not convinced that source Z is important. Where I think some confusion may have come from is Hewlett Packard's 12 term error correction analysis derived for vector network analyzers. Here source Z is important because measurements are made in both directions. I have some conceptual problems with standing waves, and reflected power, although I know that the solution to the wave equation shows a forward and reverse traveling wave. Both with uniform plane waves, and in wire transmission lines. Transmission lines can also be analyzed as a simple passive network without regard to "Reflected power". I am sure you will rip my comments to shreds, that's ok, as I may learn something. 73, Frank "Richard Clark" wrote in message ... On Tue, 07 Dec 2004 13:02:01 GMT, "Frank" wrote: What is the source Z of a solid state power amplifier HI Frank, Commonly 1.5 to 3 Ohms resistive transformed to 35 Ohms to 70 Ohms at the Connector. or even a tube amplifier? Much greater variation here, X KOhms resistive transformed to 50 Ohms at the connector. Such are ballpark figures, as an average over a full cycle, at rated power, for Class AB operation in a Push-Pull configuration. This typically results in an efficiency on the order of 40% to 60%. The nut of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then almost all their power would be reflected back into them at the antenna connector's connection to a 50 Ohm antenna system. The argument of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then reflections from the load (once you got some power into line) would encounter this same massive mismatch and re-reflect. There is a naive argument here (all too common) that this is "exactly" what happens. My pointed observation to those statements is "why would anyone need a tuner then?" The refinement of the matter about "Source Z matters" is that if your source were at either of those untransformed Zs that are native to transistors or tubes, then with any mismatch at the load you haven't got a clue what power is being applied OR reflected. This is called the Mismatch Uncertainty. It is another indicator of the failure of the argument mentioned in the previous paragraph. The relevance of the matter about "Source Z matters" here is that the references that Bob used to measure and model line loss supports this thesis. It presents an opportunity to observe how a line would suffer additional loss through being mismatched at both ends. In this regard, it would be due to the fictive argument for Source Z being very much lower or very much higher than 50 Ohms (in other words, lacking the transform circuitry commonly found in commercial gear). When the loss is not observed, the fiction is shown. 73's Richard Clark, KB7QHC |
Richard Clark wrote: This is called the Mismatch Uncertainty. The relevance of the matter about "Source Z matters" here is that the references that Bob used to measure and model line loss supports this thesis. It presents an opportunity to observe how a line would suffer additional loss through being mismatched at both ends. In this regard, it would be due to the fictive argument for Source Z being very much lower or very much higher than 50 Ohms (in other words, lacking the transform circuitry commonly found in commercial gear). When the loss is not observed, the fiction is shown. "It" should be called Grammatical Uncertainty. 73, AC6XG |
On Tue, 07 Dec 2004 18:57:25 GMT, "Frank"
wrote: "We use a matching network and adjust it for the required output power, then measure the input impedance of the network. The complex conjugate of this impedance is then defined as the source Z". The fact is these data are not the actual source Z of the device, but are probably considerable higher. I don't remember anybody actually trying to measure the large signal S parameters of solid state devices. Hi Frank, I've heard variations of this before, and other's admonitions that Motorola admitted to a huge specification mistake in the early 90s and had since mended their ways. When I asked for these updated references, I ended up quoting verbatim from those mended teachings, that, yes, Source Z has always been what was specified before (...1990s), it was the same then (1990s), and it is the same now (1990s...). I've just returned from a nanotech seminar this afternoon whose subject was organic thin film transistors. Dr. Daniel Frisbie - Depts. of Chemical Engineering and Mat. Science - University of Minnesota, offered that this new generation of research confirmed that the Resistance of the transistor channel (similar to a MOSFET) easily dominated all other sources of impedance. They also tested for junction offsets (valence band - conduction band potentials) and found they were negligible. The electron mobility wasn't the hottest thing going (semiconducting carbon nanotubes easily dominate), but there were no surprises. One of the EEs in the crowd easily allowed the OTFTs showed no chemical/physical/electrical departures from expectations (except for his concern for Schottky bias). The technique you describe above is called a transfer standard. Unless there is some mysterious shift in the space-time continuum to account for this operation being invalid, it fully and accurately describes the unit under test. Most arguments that lean on this indirect measure being suspect would have us counting electrons instead of using an Ammeter. Then we would argue about the counter and its incapability of being a direct measure, but simply another abstraction. Inevitably the arguments spiral down to the retort "you are not going to change my mind." I am already in the middle of the science that does real electron counting, literally, where one can find what is called the Coulomb barrier. For carbon nanotubes, things are so small that one electron in a "wire" cannot allow another in with it to share the conductor. Nothing like that is going on in our rigs. 73's Richard Clark, KB7QHC |
Hi Richard, thanks for your comments. My contacts with Motorola were in the
late 80s, so does put it in the correct time frame, and does not surprise me. I know I thought of it as a not very elegant method. The particular device I was thinking of was a dual push-pull (could have been parallel) module designed for about 200 - 500 MHz, at about 50 W. As far as I am concerned Motorola has gone down hill since they used to produce those thick RF device data books. Not to mention 4DTV. During that same period everybody was using a technique known as "Load-pull" for power amplifier design. I was also involved, though not very deeply, in the design of 100 W to 1 kW HF solid state linear amps, where I was told the same story about not attempting to actually match the bipolar devices, but simply present an appropriate impedance to obtain the output power; otherwise the device parameters will be exceeded. I am far from an expert in the field of power amplifier design, but it would be interesting to know if high power transistors are now characterized by large signal S parameters. This may sound really dumb, but how about feeding 100 W back into a transistor amplifier, and measuring the return loss. It would at least give you the large signal magnitude of S22. You are starting to loose me when it comes to semi-conductor physics, as it was not a field that I was especially interested in. I think the only journals that I may have read about "nanotubes" may have been Scientific American. I have also never had any experience with power FETs. 73, Frank "Richard Clark" wrote in message ... On Tue, 07 Dec 2004 18:57:25 GMT, "Frank" wrote: "We use a matching network and adjust it for the required output power, then measure the input impedance of the network. The complex conjugate of this impedance is then defined as the source Z". The fact is these data are not the actual source Z of the device, but are probably considerable higher. I don't remember anybody actually trying to measure the large signal S parameters of solid state devices. Hi Frank, I've heard variations of this before, and other's admonitions that Motorola admitted to a huge specification mistake in the early 90s and had since mended their ways. When I asked for these updated references, I ended up quoting verbatim from those mended teachings, that, yes, Source Z has always been what was specified before (...1990s), it was the same then (1990s), and it is the same now (1990s...). I've just returned from a nanotech seminar this afternoon whose subject was organic thin film transistors. Dr. Daniel Frisbie - Depts. of Chemical Engineering and Mat. Science - University of Minnesota, offered that this new generation of research confirmed that the Resistance of the transistor channel (similar to a MOSFET) easily dominated all other sources of impedance. They also tested for junction offsets (valence band - conduction band potentials) and found they were negligible. The electron mobility wasn't the hottest thing going (semiconducting carbon nanotubes easily dominate), but there were no surprises. One of the EEs in the crowd easily allowed the OTFTs showed no chemical/physical/electrical departures from expectations (except for his concern for Schottky bias). The technique you describe above is called a transfer standard. Unless there is some mysterious shift in the space-time continuum to account for this operation being invalid, it fully and accurately describes the unit under test. Most arguments that lean on this indirect measure being suspect would have us counting electrons instead of using an Ammeter. Then we would argue about the counter and its incapability of being a direct measure, but simply another abstraction. Inevitably the arguments spiral down to the retort "you are not going to change my mind." I am already in the middle of the science that does real electron counting, literally, where one can find what is called the Coulomb barrier. For carbon nanotubes, things are so small that one electron in a "wire" cannot allow another in with it to share the conductor. Nothing like that is going on in our rigs. 73's Richard Clark, KB7QHC |
On Wed, 08 Dec 2004 01:25:07 GMT, "Frank"
wrote: I was told the same story about not attempting to actually match the bipolar devices, but simply present an appropriate impedance to obtain the output power; otherwise the device parameters will be exceeded. Hi Frank, There used to be an old, old song: "Yes dear you can go swimming, but don't go near the water." Even at DC, you cannot design to the capacity of a transistor, because the combination of all capacities exceed the "safe operating area." However, this does nothing to actually change any noted specification. I've never seen ANY power amplifier for retail trade conjugately matched to its Source Z. But then I have never seen ANY amplifier for retail trade designed to be low noise, low distortion, high stability, or any of the more common qualities that "could be" designed in, if it weren't for cost and the perception of no particular boon to the purchaser. Who needed low distortion when you could throw a cheap filter on the output? Who need low noise when atmospherics dominated such issues? Who needed stability when the monkey twisting the knob would correct it as a form of entertainment? Safety margin? Add a fan to the heatsink - $pecial option. This may sound really dumb, but how about feeding 100 W back into a transistor amplifier, and measuring the return loss. It would at least give you the large signal magnitude of S22. This was done decades ago - it is called an active load. I used to calibrate them too. Guess what, it was specified and it met spec at 50 Ohms (and at least 100W). 73's Richard Clark, KB7QHC |
Frank wrote:
If you remember, Motorola used to publish Smith charts of the output impedance for their power amplifier devices. Talking to one of Motorola's design engineers; I asked "How do you derive these Charts". His answer was; "We use a matching network and adjust it for the required output power, then measure the input impedance of the network. The complex conjugate of this impedance is then defined as the source Z". The fact is these data are not the actual source Z of the device, I had heard that also. For a typical VHF/UHF device, the manufacturer's application engineers use an infinitely adjustable stub tuner to explore the whole range of possible load impedances presented TO the device. As well as measuring output power, the application engineer also has to think about maximum voltage and current ratings, chip and bond wire temperatures, and also IMD performance if the device is going to be specified for linear operation. The application engineer adjusts the load impedance to give the optimum balance of all these factors, at a series of test frequencies. No problems whatever about that. The only technical issue is the *assumption* that the conjugate of the load impedance is equal to the output impedance of the device. Most manufacturers now tend to avoid that assumption, because it is a totally unnecessary distraction for the transmitter designer who has to use the device. All the designer has to do is create an output network that presents the manufacturer's recommended load impedance TO the device. This network replicates the impedance transformation of the original stub tuner setup, but uses mostly fixed components for obvious practical reasons. Apart from a very few special applications where reverse termination is important to avoid ghosting and similar effects, the transmitter designer doesn't have to think about the device's output impedance at all. -- 73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB) http://www.ifwtech.co.uk/g3sek |
Ian White, G3SEK wrote:
"The only technical issue is the "assumption" that the conjugate of the load impedance is equal to the output impedance of the device." It`s true if maximum power is being transferred. King, Mimno, and Wing sat so on page 43 of "Transmission Lines, Antennas, and Wave Guides": "If a dissipationless network is inserted betweeen a constant-voltage generator of internal impedance Zg, and a load of impedance ZR such that maximum power is delivered to the load, at every pair of terminals the impedances looking in opposite directions are conjugates of each other." The authors, Arnie, Larry, and Alex were all teaching at Harvard in 1945 when their book was published. Walter Maxwell, W2DU has been saying the same thing yet has mistaken nay-sayerrs. Best regards, Richard Harrison, KB5WZI |
"Richard Harrison" wrote in message
... Ian White, G3SEK wrote: "The only technical issue is the "assumption" that the conjugate of the load impedance is equal to the output impedance of the device." It`s true if maximum power is being transferred. King, Mimno, and Wing sat so on page 43 of "Transmission Lines, Antennas, and Wave Guides": "If a dissipationless network is inserted betweeen a constant-voltage generator of internal impedance Zg, and a load of impedance ZR such that maximum power is delivered to the load, at every pair of terminals the impedances looking in opposite directions are conjugates of each other." The authors, Arnie, Larry, and Alex were all teaching at Harvard in 1945 when their book was published. Walter Maxwell, W2DU has been saying the same thing yet has mistaken nay-sayerrs. Best regards, Richard Harrison, KB5WZI Maximum power transfer with conjugate matching is undisputed. The problem with semi-conductor devices is that you cannot necessarily conjugate match because the device operating parameters may be exceeded. 73, Frank |
Frank wrote:
"Richard Harrison" wrote in message ... Ian White, G3SEK wrote: "The only technical issue is the "assumption" that the conjugate of the load impedance is equal to the output impedance of the device." It`s true if maximum power is being transferred. King, Mimno, and Wing sat so on page 43 of "Transmission Lines, Antennas, and Wave Guides": "If a dissipationless network is inserted betweeen a constant-voltage generator of internal impedance Zg, and a load of impedance ZR such that maximum power is delivered to the load, at every pair of terminals the impedances looking in opposite directions are conjugates of each other." The authors, Arnie, Larry, and Alex were all teaching at Harvard in 1945 when their book was published. Walter Maxwell, W2DU has been saying the same thing yet has mistaken nay-sayerrs. Best regards, Richard Harrison, KB5WZI Maximum power transfer with conjugate matching is undisputed. The problem with semi-conductor devices is that you cannot necessarily conjugate match because the device operating parameters may be exceeded. Exactly. Richard's assertion relies on at least three things being true: 1. That maximum power is being transferred - for most states of transmitter tuning, loading and drive levels, that is obviously *not* true. 2. That the transmitter can be accurately represented as a "constant-voltage generator of internal impedance Zg", i.e. as a Thevenin source. 3. That as part of #2, Zg is a constant. With all due respect to Richard - and above all, respect to Walt - it is a tall order to prove that all three of those requirements for conjugate matching are being met. I believe they can only be exactly met under a few very special sets of operating conditions. But it then follows that, for all *other* operating conditions, the complete end-to-end conjugate matching referred to by King et al does *not* exist. -- 73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB) http://www.ifwtech.co.uk/g3sek |
"Ian White, G3SEK" wrote in message
... Frank wrote: If you remember, Motorola used to publish Smith charts of the output impedance for their power amplifier devices. Talking to one of Motorola's design engineers; I asked "How do you derive these Charts". His answer was; "We use a matching network and adjust it for the required output power, then measure the input impedance of the network. The complex conjugate of this impedance is then defined as the source Z". The fact is these data are not the actual source Z of the device, I had heard that also. For a typical VHF/UHF device, the manufacturer's application engineers use an infinitely adjustable stub tuner to explore the whole range of possible load impedances presented TO the device. As well as measuring output power, the application engineer also has to think about maximum voltage and current ratings, chip and bond wire temperatures, and also IMD performance if the device is going to be specified for linear operation. The application engineer adjusts the load impedance to give the optimum balance of all these factors, at a series of test frequencies. No problems whatever about that. The only technical issue is the *assumption* that the conjugate of the load impedance is equal to the output impedance of the device. Most manufacturers now tend to avoid that assumption, because it is a totally unnecessary distraction for the transmitter designer who has to use the device. All the designer has to do is create an output network that presents the manufacturer's recommended load impedance TO the device. This network replicates the impedance transformation of the original stub tuner setup, but uses mostly fixed components for obvious practical reasons. Apart from a very few special applications where reverse termination is important to avoid ghosting and similar effects, the transmitter designer doesn't have to think about the device's output impedance at all. -- 73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB) http://www.ifwtech.co.uk/g3sek It seems everybody is in agreement with the fact that you cannot match a power source to the load. As Ian states: "nobody cares, what the device output parameters are, only that it is capable of delivering the required power to a desired load". This still leaves the question unanswered as to "What is the actual device S22"? I have read that "With HF linear devices the large signal S parameters are close enough to the small signal values". For non-linear devices load-pull techniques are used. I have never seen high power transistors characterized with S parameters, but have not worked with such designs for a number of years, so am probably out of touch. A tentative search of the web did not find any info. It seems TRW and Motorola are pretty much out of the power semi-conductor industry. I am tempted to synthesize a transmission line based on the per-unit length parameters, and see how the load power varies as a function of source Z. It seems to me the only important factor is the applied voltage. The input impedance is only a complex number. The transmission line could be considered a "Singly terminated network", the synthesis of which is trivial, and performance independent of source Z. I have trouble with the concept of "Reflection"; how can charges (electrons) flow in both directions simultaneously. Charge flow results from the E field within the conductor. 73, Frank |
On Tue, 07 Dec 2004 18:57:25 GMT, "Frank"
wrote: |Hi Richard, | |With solid state power amplifier design; the criteria was always that you |must present an impedance, to the output devices, such that the desired |output power is delivered to the load (while not exceeding device |dissipation). Any attempt to optimally match the load to the source |impedance will result in over-dissipation, and probable destruction of the |source device -- probably by excess collector/drain current. If you |remember, Motorola used to publish Smith charts of the output impedance for |their power amplifier devices. Talking to one of Motorola's design |engineers; I asked "How do you derive these Charts". His answer was; "We |use a matching network and adjust it for the required output power, then |measure the input impedance of the network. The complex conjugate of this |impedance is then defined as the source Z". The fact is these data are not |the actual source Z of the device, but are probably considerable higher. The data were useful as presented. When using a Smith chart for matching network design, the published data could be used as the "starting point" for the network and the rotations were made toward the load; the opposite from the usual case of matching a load to a 50 ohm source. This method was really an early example of load pull characterization. Maury Microwave app. note 5C-041 is one reference for this. Another is "A New Load Pull Measurement Technique Eases GaAs Characterization", Microwave Journal, Nov, 1980, pp. 63-67. |I don't remember anybody actually trying to measure the large signal S |parameters of solid state devices. |I seem to remember that tube amplifiers were designed based on the source |impedance calculated as 2Vp/Ip, (Where Vp is the plate voltage, and Ip the |plate current), and have no idea how, or if, it relates to the actual source |Z of the device. |Anyway, I am not convinced that source Z is important. |Where I think some confusion may have come from is Hewlett Packard's 12 term |error correction analysis derived for vector network analyzers. Here source |Z is important because measurements are made in both directions. It's not the error correction that is confusing. The error correction simply removes systematic errors from the measurement(s). The parameters usually measured when 12-term correction is called for are the small-signal S-parameters. If I understand you correctly, "source Z" is actually the output reflection coefficient (s22), the signal exiting port 2 due to an input to port 2. The amplifier output reflection coefficient can be very important, even in high power amplifiers, when non-dissipative filters are used for harmonic or spurious rejection. Such filters are commonly specified and measured in 50 ohm systems and function by reflecting, not dissipating, the out-of-band energy. When driven by other than a 50 ohm source the rejection will be other than what is measured in a matched condition. |
Frank wrote:
I have trouble with the concept of "Reflection"; how can charges (electrons) flow in both directions simultaneously. Have you ever stood on a cliff overlooking the ocean and seen ocean waves rolling in and smaller waves rolling back out? The smaller waves rolling back out to sea are reflections of the large waves incident upon the shore. The small outflowing wave meets a large incoming wave and seems to disappear, only to emerge on the ocean side of the large wave with its identity still intact. If ocean waves can flow both directions using the same H2O carriers, why would anyone have difficulty in accepting EM waves flowing in both directions using the same electron carriers? The energy in the ocean waves travels much faster than the water molecules. The energy in an EM wave travels much faster than the electrons. Ever played with a long rope fastened at one end? You can send a wave down the rope and receive a reflected wave. If you time it just right, you can have a forward wave and a reflected wave meet in the middle of the rope and be unaffected by each other as long as things remain linear. A forward EM wave has no effect on a reflected EM wave and vice versa as long as things remain linear. -- 73, Cecil http://www.qsl.net/w5dxp |
On Wed, 08 Dec 2004 15:51:07 GMT, "Frank"
wrote: Maximum power transfer with conjugate matching is undisputed. The problem with semi-conductor devices is that you cannot necessarily conjugate match because the device operating parameters may be exceeded. Hi Frank, If we return this from the ethereal landscape of sub meter wavelengths, back to the point of Bob's measurements at HF, and lately the MF; then matching and issues of the final are trivial with a million examples in the market today. I know you would probably like to get to the nut of this, and it will return you to Motorola's AN1526. This work contains much of the language offered by correspondents here (in times past), but through the rather gauzy filter of their memory. Usually they couch the disassociation of Source Z to a transistor through poor context (in other words, not reading the entire subject, but just a phrase). There is the presumption these posters embrace: "They consider the best match is achieved by a simultaneous conjugate match of the input and output. However, power amplifiers provide higher power gain and better efficiency at the rated output power if the output is purposely mismatched. An added benefit of doing this is potentially unstable devices, conjugately matched, can be operated stably under these more optimum mismatched conditions." This is the usual mistake of misattribution between the distinctions of a Conjugate Match, and a Z Match. However, you will note that here, and elsewhere in the reference, that no one denies the Source has a Z, and it is significant (all within values I've offered) and that it is still closely held to the expected load (later I will show exactly held). Another dismissal offered by posters is the supposed invalidity or inaccuracy of the load-pull method (which I find curious after having calibrated active loads suited for just this purpose). I will turn again to this same reference: "Although the technique has been known for some time, the widespread availability of desktop computers and automatic tuning systems is just now making this method more attractive, particularly for higher power devices. The characterization process is conceptually quite simple." Then there is the subject of S parameters, which is introduced early by Motorola with this admonition: "Many first time RF power designers, brought up on a diet of small–signal s–parameters, previously used for solving small signal text book problems, assume these same techniques are applicable to bipolar class–C and class–AB power amplifier design." This selection actually introduces the presumption above. We have one poster here that violates this admonition with abandon - but with regard to transmission lines. When the poster pines further for a Large Signal S parameters: "However, the authors are not aware of these parameters being used successfully above a few watts of output power." When Motorola actually gets down to design: "The load line resistance is the optimum load impedance for the internal collector node of the transistor, neglecting the junction and parasitic device capacitance." What a concept! Same as before, Same then, Same now. The ONLY contretemps revealed by this tempest in a teapot is the forced conclusion that a conjugate match was required (a common mistake of not knowing the difference between Conjugate Matching and Z Matching) which was in turn driven by higher frequency operation (much of this is couched in the 900 MHz band) and parasitics already noted above. The final and most compelling admission from Motorola is found with their statement: "It is up to the device designer to choose which impedance gets published. One is just as valid as the other. However, quite frankly, gain is what sells devices." And of course this discussion will do nothing with those who utter "you are not going to change my mind." ;-) To be continued - no doubt. 73's Richard Clark, KB7QHC |
Frank, WE6CB wrote:
"Maximum power transfer with conjugate matching is undisputed." Great! W2DU has made progress. Frank also wrote: "The problem with semi-conductor devices is that you cannot necessarily conjugate match because the device operating parameters may be exceeded." True for certain voltages, currents, and drive. The maxima may not all be acheived simultaneously. It`s true for vacuum tube amplifiers too. The total device dissipation can`t be exceeded without reduced life expectancy. Another caution is the difinition of "maximum available power". In an RF amplifier this is specific to a certain set of operating conditions. Maxium available power is with fixed B+ (or minus) voltage and drive. Current, too, is limited to non-destructive values. Terman says on page 76 of his 1955 edition: "Alternatively, a load impedance may be matched to a source of power in such a way as to make the power delivered to the load a maximum. (The power delivered to the load under these conditions is termed the avalable power of the power source.)" Best regards, Richard Harrison, KB5WZI |
Frank wrote:
"I have trouble with the concept of "Reflection", how can charges (electrons) flow in both directions simultaneously." The wave, or signal flowing in one direction is distinctly different from that flowing in the opposite direction. Upon reflection, the phase between current and voiltage prodiuced by the wave is inverted. Only phase of the volts or amps is inverted by reflection, not both. The phase relation between volts and amps is the key to the direction the wave is traveling on the line. It`s the wave which travels. The volts and amps are generated by the wave traveling on the line. The line is just guiding the wave. The traveling forward and reflected waves, traveling in opposite directions on the same line, produce the familiar standing wave patterns through superposition of volts and amps. Transmission lines and the appurtenances used with them have no problem keeping values associated with the waves straight with proper design. A directional wattmeter can separate the two directions of travel wery well indeed. It knows one direction from the other by whether the volts and amps are in-phase or out-of-phase. Best regards, Richard Harrison, KB5WZI |
I quoted Terman, saying:
"The power delivered to the load under these conditions (a conjugate match) is termed the AVAILABLE POWER of the power source." The match between source and load is the best it can be and can`t be improved when you have a conjugate match. A conjugate match is an empowerment but does not cause you to put out any particular power. That`s up to you. Suppose you have a conjugate output match to a Class-B power amplifier you are driving with your SSB transmitter. Instantaneously, average power from the amplifier is following the modulation. A single steady tone ideally produces a particular output at at one radio frequency. Want more output? Increase drive to the amplifier. Want less? Reduce drive to the amplifier. You may have a conjugate match under only one condition, some conditions, or under all conditions, but given life`s usuall imperfections, I would place no bets, except against all conditions. Best regards, Richard Harrison, KB5WZI |
On Wed, 8 Dec 2004 20:55:18 +0000 (UTC), "Reg Edwards"
wrote: and if that isn't enough, to further complicate matters, the internal impedance of the transmitter changes as the load impedance is varied Hi Reggie, Such arguments are as juvenile as the claim no one can travel a straight line because the earth is rotating under them. A quadrillion miles of experience would suggest this too is trivial to accomplish. Aren't you the one who is so charmed with the legacy of Kelvinator who chimed that such chimera without calculation are the chatter of chimps in the forest canopy? 73's Richard Clark, KB7QHC |
Just a comment -
The design, from start to finish, of a linear power amplifier is based solely on a device's ratings - volts, amps, watts, etc. Its RF internal impedance plays no part in it. At HF it is never specified by the manufacturer. Even ARRL bibles don't mention the subject of Rint. It's superfluous. Does anybody know what it is? Give us some numbers. As for conjugate matching - don't make me laugh. --- Reg. |
Richard, you've slipped into one of your incoherent phases. Try again in a few days time when you are feeling better. ;o) ---- Yours, Punchinello. |
Reg, G4FGQ wrote:
"The active device generally behaves as a current source." As Reg also wrote: "I can`t imagine why this conversation has continued for so many years by more or less the same group of experts." Agreed! Reg seems to have answered his own question.The same people recite the same arguments in hopes their view of reality will be accepted. Fat chance! Time has inured them. Reg has faithfully proposed constant-current behaviour from all vacuum valves and transistors as I recall. I agree that most of these devices have extremely high plate ond collector resistances as linear amplifiers. Current through them is almost constant regardless of anode voltage. As most transmitter power amplifiers exceed 50% efficiency by a good margin, these devices are not operating as Class-A linear amplifiers. They instead operate as HF switches. These are turned-off most of every cycle and are only on for short pulses. Harmonics and other noise is cleaned up by output filters. It`s the only thing which makes the output linear. During the output device`s conduction, its saturation volts are very low and its current is very high, giving the device a very low impedance while switched-on. You may not infer a low impedance from the d-c volts and amps feeding the final amplifier. These are the averages, almost, of the device amps. The device saturation volts sre what counts toward its dissipation and loss. The transmitter usually has no built-in indicator of saturation voltage. It wouldn`t read much anyway.Device impedance depends mostly on its ratio of off to on times. This is a form of lossless resistance. Dissipation is zero in a sewitched-off device. The d-c volts and amps are related to the output device(s) internal impedances used as a switch when the transmitter output is considered. A high voltage and a low current accompany a high internal impedance but they won`t be nearly so high as the spec sheet plate or collector resistances. We have d-c power input to the amplifier. We can measure HF power output. The difference is dissipation, but loss resistance does not represent the total source resistance because we have non-dissipative resistance in the device off-times. There have been measurements of transmitter internal output impedances which indicated that they did indeed match their loads. I have not done it myself but have no reason to doubt the reports. Best regards, Richard Harrison, KB5WZI |
| All times are GMT +1. The time now is 09:06 PM. |
Powered by vBulletin® Copyright ©2000 - 2025, Jelsoft Enterprises Ltd.
RadioBanter.com