Dave, G3UUR sent me his last write-up on the Conjugate Match debate today where he was finally able to disprove that claim. Attached below is the last full unedited document. An older slimmed down version appeared in AntenneX magazine a few years back. This became a very hot and often heated topic on the ham boards and peaked around 2011. Dave then got busy with other stuff and has been absent for the past few years. The below was copied and pasted from the last part of his attached pdf document and a few of the symbols may not have transferred so refer to the document for anything that might be missing. Dave Gordon-Smith, G3UUR: New Method of Measuring ROS ……… 27th July 2011 Conclusions The measurement of amplifier output resistance described in this article shows conclusively that there is a loss mechanism associated with tube operation that is referred to the output of any power amplifier as a dissipative resistance. This demonstrates quite clearly that ROS is not a non-dissipative resistance as W2DU and VE2CV have inferred from their limited evidence, because such a resistance would not broaden the bandwidth of any network following the amplifier. In addition, the magnitude of ROS is substantially less than the load resistance, RLOAD, even when the amplifier is tuned for maximum output according to the reverse SWR (RPG) method. This is because the tube source resistance, RDS, and optimum load, RL, are governed by two totally different tube parameters. In this respect a tube power amplifier is quite different to an idealised Thevenin source, and trying to interpret experimental results using a model based on the latter will always lead to erroneous conclusions. The RPG experiments conducted by W8JI  and the load-variation measurements made by W2DU  are only capable of detecting the time-averaged value of the dynamic plate resistance, rp. It is this parameter that determines the optimum value of load for maximum power transfer and the resistance that terminates the input of the tank circuit for any probe signal being fed back into the amplifier during a reverse SWR test. Plate resistance is just the name given to the mathematical ratio Ep/Ip with Eg constant. It’s a measure of how much varying the plate voltage influences the plate current. It affects the way the power delivered to the load changes as the load line is altered and behaves just like RS in a Thevenin source in this respect, causing a peak in output power and an ‘apparent’ conjugate match. However, it’s not a real resistance and does not dissipate any power when the tube conducts because none of the plate current due to grid drive goes through it. So, it cannot provide the input termination for any selective network or antenna following the amplifier. Only dissipative resistance can do that. Doug Smith, KF6DX, has already pointed out  the inconsistency between the results of RPG measurements and the claim that the source resistance is nondissipating, but to date no credible explanation has been given by the pro-CM lobby to resolve this difficulty. In reality, of course, the probe signal in an RPG experiment sees the plate-cathode resistance disguised as rp for the time the tube conducts because it only varies the plate voltage. This terminates the input of the pi-network tank circuit for any reverse signal and presents it with a radically different value to the loss resistance experienced by the plate current driving the tank circuit. The probe signal alters the plate current asynchronously by a small amount and is dissipated at the plate along with the power that’s lost from the flow of plate current that’s going into the tank circuit. Plate resistance is the only characteristic of a tube that can be non-dissipating in the forward direction and appear to be dissipating in the reverse. No other characteristic of the tube or matching circuit can behave in this way. The total dissipative resistance experienced by the plate current is governed by the sum of all the instantaneous power lost across the tube during the time it’s conducting. Part of this lost power is associated with wave shaping (RWS) and part of it with the output resistance (RDS). The latter is determined by epmin/ipmax and therefore related to the drive level and the proportion of the DC supply voltage used by the plate swing. This has been confirmed by tests on linear amplifiers where the drive and plate swing could be varied. So, other properties of the tube, and not rp, determine the level of efficiency and the dissipative output resistance. The whole operation of a tube HFTPA is not just subtly different to a Thevenin equivalent source; it’s radically different to it because all 3 properties associated with RS in a Thevenin source are governed by different characteristics of the tube in a power amplifier. Therefore, the Maximum Power-Transfer Theorem, which is based on a derivation involving a Thevenin equivalent source, cannot be applied to power generators using active devices. However, the conjugate match is still a useful concept when considering matching problems and does apply to other power sources such as receiving antennas, so should not be dismissed as a theoretical irrelevance and ignored. It’s worth noting that the peak in output observed during RPG and loadvariation tests on AB1 amplifiers occurs at power levels higher than those recommended for linear operation. This is because the value of Rp, and hence rp, in real tubes varies across the plate characteristics, and is normally too high to meet the optimum-load condition at moderate and high plate voltages, though load-variation measurements still show an apparent source impedance of 60 to 80 under these conditions because of the diluting and obscuring effect of the pi-network transformation. For load-variation figures closer to 50, the end of the load line has to be pushed further down into the non-linear, low-voltage, high-current region of the plate characteristics by increasing the value of the load presented to the tube by the pinetwork matching circuit. Then, rp can be forced to drop significantly so the optimum-load condition can be met by some lower effective value averaged over the plate swing. It is only in this condition that the reverse SWR method provides a perfect match well over the recommended output for tube linear amplifiers. Perhaps Warren Bruene  had it right when he said “maximum power output is clearly determined by non-linear tube conditions.” New results presented by W2DU  in Sec 19A.5 actually support the idea that the peak in power at maximum output occurs because of the interaction between the load and the tube plate resistance, rp. Walt doesn’t agree, of course. However, his interpretation of these results ignores the fact they show quite clearly that the tank circuit is transparent to steady-state changes and he has an ‘apparent’ conjugate match right at the plate where he’s making the measurements. By suggesting that the Q somehow isolates the input of the tank circuit from the output, he’s blatantly ignoring the fact that he can measure this ‘apparent’ conjugate match right at input of the tank circuit. How there can be any isolation at all if he can do that is very hard to believe! He’s also breaking the golden rule of transmission-line theory, which is that if a conjugate match occurs anywhere in a system it occurs everywhere, including right back to the source – this should be mentioned somewhere in one of the earlier chapters of his book. His claim also ignores a basic tenet of circuit theory that the dissipative resistance presented to the input of a network is always referred through to the output in some form, or other, depending on its relationship to the characteristic impedance and the phase delay of the network, regardless of any energy storage along the way. Various types of filter with Q values far in excess of the tank circuit in an HFTPA refer the dissipative parts of their input and output terminating impedances through to the other port. This is a well established fact and tank circuits with only modest values of Q are no exception. If rp was a dissipative resistance rather than a mathematical ratio, there would be a conjugate match at the plate, but it doesn’t and there isn’t. Correcting Misconceptions After the publication of W8JI’s RPG test results  in 1997, there seems to have been a gradual acceptance amongst radio amateurs that a conjugate match exists in RF power amplifiers and the myth has become well established now. However, this widespread acceptance appears to have come with little recognition that it requires the output impedance to be partially non-dissipating to explain the more than 50% efficiency obtained with most tube RF power amplifiers. Some amateurs now routinely assume that the broadening effect of the output impedance makes the bandwidth twice what it would be for the load alone, whether it does or not! It seems more attention should have been paid to the pertinent points made by Warren Bruene  in a letter written to the editor of QEX back in 2005. He pointed out that a tube is basically a voltage generator, despite the fact that the plate current is grid-voltage controlled, with such a low value of output resistance that it’s possible to create an impedance inversion and get a current generator with a high value of source resistance if the tank circuit and transmission line together provide an odd number of 90º phase delays. Apparently, this is a trick discovered by broadcast engineers to increase the field-strength response of high-Q antennas to improve the transmitted bandwidth. So, in practice, the broadening effect of the dissipative output resistance of a power amplifier can be anything from the minimal influence of a voltage generator with low dissipative output resistance to the considerable widening effect of a current generator, depending on the phase delay of the matching network and the length of the transmission line. This would not be the case if a conjugate match were a reality in RF amplifiers because a line with very low loss would be matched fairly well looking down it in either direction at every point along it, no matter what its length, and no variation of the effective source impedance would be evident. The conjugate match in HF tuned power amplifiers is not only elusive, it’s an illusion! References 1 M. Walter Maxwell, W2DU, Chapter 19A in “Reflections III: Transmission lines and Antennas,” published by CQ Communications Inc., April 2010, ISBN: 0943016436. 2 Walter Maxwell, W2DU,”On the Nature of the Source of Power in Class-B and –C Amplifiers,” QEX, May/June 2001, pp32 - 44. 3 Robert Craiglow, “RF Power Amplifier Output Impedance Revisited, QEX, Jan/Feb 2005, pp 29-37. 4 Warren Bruene, W5OLY, “The Elusive Conjugate Match,” Communications Quarterly, Spring 1998, pp 23-31. 5 Jack Belrose, VE2CV, Walter Maxwell, W2DU, and Tom Rauch, W8JI, “Source Impedance of HF Tuned Power Amplifiers and the Conjugate Match, “Communications Quarterly, Fall 1997, pp25 - 40. 6 Douglas Smith, KF6DX, “Letters to the Editor,” QEX, May/June 2006, p61. 7 Warren Bruene, W5OLY,”Letters to the Editor,” QEX, Mar/Apr. 2005, p62.