Simulating tube amplifiers

Discussion in 'Amateur Radio Amplifiers' started by W8JI, Dec 29, 2011.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: Left-3
ad: Left-2
ad: MessiPaoloni-1
ad: Subscribe
ad: L-MFJ
  1. G0HZU

    G0HZU QRZ Member

    To get your head around what I am getting at here, the structure is a vertical column and I'm doing multiple simulations stacked on an ever growing inductive column.

    Because the loop gain goes up and phase changes as you climb the structure then or two things happen. You either get part of it oscillating (the top bit)
    Or if it has strong feedback then more cells lower down kick off as well but at a slightly higher frequency. So you get a chaotic chorus of electrons?

    Is this feasible? I don't really do particle physics so I run out of road at this point...

    But my guess is you can't treat this oscillation condition the same as regular amplification. I think there will be multiple events within parts of the structure spread over several MHz generating high RF voltages if the phases all combine at the common anode cap.
    Last edited: Jan 3, 2012
  2. G0HZU

    G0HZU QRZ Member

    What the big model shows is a change in loop gain (AT VHF) of about 8dB across the height of the structure. i.e. the top part of the column has 8dB more LOOP gain than the bottom. This isn't the same as just measuring valve gain. This is gain around the feedback path.

    Also the zero phase crossing point shifts about 20MHz around the path according to column height.

    So what may happen is that some people may see their amp hoot at VHF with no suppressors but it may only be the very top of the structure that has adequate loop gain for oscillation.

    So the event could be harmless apart from some RFI.

    But if a high gain tube was used in an amp with poor layout and a certain tuning/band setting then there could be a major event that means that most of the vertical structure has adequate loop gain to oscillate.

    You then get more heat across more of the structure (stress?) and also lots more scope for high combined peak voltage as there will be multiple oscillations in parallel across several MHz up and down the inner structure.
  3. W8JI

    W8JI Ham Member QRZ Page

    Personally, if I really wanted to model a tube outside the normal range where it works, I would slow down and think through the tube. Logic goes a long distance further that bouncing off the walls with wild notions.

    Even a large grid like a 3-500Z has minimal standing waves on the grid structure, because there simply isn't room for the wave to stand. At 150-200 MHz, a wavelength is almost 6 feet long. A node in a transmission line 3 inches long would not have that much current or voltage variation. Maybe you have never even looked at a grid or filament?

    If you have a coaxial structure like a tube, and open terminate the end, exactly how much voltage change along .04 wavelengths of line length?

    Outside the grid structure, in some tubes, very thin single wires connect to the outside. In some cases there might be 3 total inches of a single very thin wire, then some more thicker conductors for an inch or two, but none of it coaxial. The ends terminate in low reactances. Now we can have a large voltage gradient across the system, because it appears like a mixture of lumped components.We still can't have significant standing waves in the current, but we can in the voltage. The surge impedance of a #18 or #20 AWG wire running out through the glass is a whole lot higher than the surge impedance along a half-inch cylinder. The high series impedance of a single very-thin-compared-to-length conductor allows a lot more voltage drop with high current than a large cylinder with dozens or hundreds of parallel wires that are cross connected.

    I've never yet seen any sign of localized heating of small parts of structures caused by some weird event in one area of an element, although I'm certain it could happen of the structure was a large fraction of a wavelength long. Since a wave is 15 to 20 inches long, exactly how are we going to get localized heating of small segments?

    Now if you can get a tube to run at 1500 MHz, that idea might have some merit in a three inch long cage of cross-connected wires.

    You better revamp the lumped components to look more like a transmission line with lumped components distributed at points along the length.

    Maybe you have been misled by some pictures that claim something that was impossible to happen, without any evidence it happened?

    Look at things and picture what the are based on what they look like, not what someone else tells you they look like. Use some common sense, and ask yourself, "Could I possibly make this effect appear with the claimed stimuli?". You'll see, many times, a great contradiction between what is claimed in a caption and what can logically happen even if ideally forced to happen.

    Back on the topic of the model. I think a two terminal negative resistance could be used to represent a particular system at a particular range of frequencies. That would require some stability testing of the system. The problem is, it would then only apply to that system.

    What we would find is a dozen ways to skin the cat, all of which work. Then when the next system comes along, we have to do the tests and make another different model. If I was interested in making my job take about 20 times longer to accomplish the same end-result as a day of cut-and-try, I might have done that myself years ago.

    73 Tom
    Last edited: Jan 4, 2012
  4. AG6K

    AG6K QRZ Member QRZ Page

     One would go from a current max to a current min in a quarter wavelength, and at 200MHz, it's not that far - especially when one adds C to the HV point.

     However, according to Fred Terman, 6FT, the difference in L between 5 mil diameter conductor and 500-mil diameter conductor is less than 100 x as much. Example: conductor length 2-inches: for 5-mil L=62nH and for 500-mil L=20nH.

     I've seen it happen to 8877 grids. The least amount of gold plating evaporates off the top of the grid and at the grounded end so much gold has evaporated that patches of the molybdneum base metal can be seen. In order to do this amazing feat it would take a large amount of UHF current.
    • Rich, ag6k

  5. AA5CT

    AA5CT Ham Member QRZ Page

    Get ready for this - I located the source of the 120 MHz 'ringing' (i.e., something is showing obvious resonance).

    With the 10 M coil end disconnected from the PA Tune cap, on 80, 40, 20, and 10 but not 15 the balance of the Pi tank has resonance around 120 MHz. Probably due to other coils forming self-resonance when shorted. This shows up in the plate circuit too, as observed earlier. What significance this has, yet to be shown shown ...

    Maybe this is part of what guys have seen in the past, and comes into play in different tuning and antenna load scenarios (besides some other 'funnies' in the SB-220 et al incl. the funny grid circuitry).

    Work progresses ...

    Jim de WB5WPA
  6. AA5CT

    AA5CT Ham Member QRZ Page

    Original post: Rescinded by author.
    Last edited: Jan 4, 2012
  7. G0HZU

    G0HZU QRZ Member

    I don't think you appreciate how simple this concept is. You can dismiss it as a wild notion and you may be correct but I've seen the model running and looked at the different bode plots for different parts of the structure.

    It cannot CHANGE the physics around the oscillation loop.

    The TOP of the structure will have more loop gain compared to the bottom.
    The TOP of the structure has a different phase plot around the loop to the bottom because it has a different relationship to chassis ground.

    So you could find that the top bit can oscillate because it has maybe 8dB more loop gain than the bottom. The bottom of the tube can't suddenly 'catch up' and oscillate in sympathy because it would defy the physics of its loop response. There isn't enough gain around the loop.

    But in a major event the whole thing could be unstable top to bottom if all parts do have enough loop gain with certain tuning settings or a high gain tube is fitted.
  8. G0HZU

    G0HZU QRZ Member

    Just to clarify, I'm just reporting what I find in the model and it may be true or false in reality. I'm fairly clueless about tubes :)

    But the model says that the top of the tube is the part that is most likey to be unstable.

    I think the key thing is that if there is a lot of loop gain across a lot of the structure then I think parts of it can be unstable at different frequencies at the same time. This is due to the structure having different phase shift around the loop according to frequency. This multi tone mode could generate high voltages at the anode.

    Note that I'm not suggesting the rest of the structure is totally dormant because it will still act like a stable amplifier and will amplify whatever gets fed back around the loop by the unstable energy.

    I hope people don't think I'm TELLING them what will definitely happen because it's just a model. However, I think the results are worth debating :)
  9. W8JI

    W8JI Ham Member QRZ Page


    You are proposing standing waves.

    What is the common mode, or series impedance, of the grid structure from end-to-end?

    What is the equivalent series inductance of that structure?

    What is the distributed capacitance along that structure?

    Unless that STRUCTURE has high series inductance in comparison to shunt capacitance to something else along the length of that structure, you can't have standing waves. What you are proposing is transmission line behavior.

    Try to make that happen at 150-200 MHz, with an air dielectric transmission line, with a center conductor about 1/2 inch diameter, and no lumped components (distributed capacitance only), in 3" or less of length.

    In 15 to 20 inches, with a thin conductor, there could be very large standing waves in a lossless line of high surge impedance at 150-200 MHz. As we make the conductor larger diameter, standing waves diminish. As we make it shorter, moving away from 1/4 wave, the standing wave diminishes. Every dimension starts to approach the length. If I wanted to build a broadband constant current and voltage termination, I'd use a very large diameter conductor compared to length.

    How are you getting away from that effect?

    73 Tom
    Last edited: Jan 4, 2012
  10. G0HZU

    G0HZU QRZ Member

    I meant there might be local heating at the microscopic level in the boundary across the focussed region that was acting like an oscillator and the bit right alongside that is still just acting like an amplifier.

    I don't know how the electrons behave in this boundary region where there could be regenerative gain in regions where instability is borderline. I don't want to give the impression I'm into particle physics. I'm really just guessing... :)
Thread Status:
Not open for further replies.

Share This Page